elimination of cross sensitivity in admittance …s2is.org/issues/v7/n4/papers/paper32.pdf ·...
TRANSCRIPT
ELIMINATION OF CROSS SENSITIVITY IN ADMITTANCE
TYPE LEVEL MEASUREMENT USING FUZZY BASED
LINEARISER
Joyanta kumarRoya and Bansari Deb Majumder
b
aM.C.K.V college of Engineering, WB, India. bNarula Institute of Technology, WB, India.
Emails: [email protected] , [email protected]
______________________________________________________________________________
Submitted: Oct. 16, 2014 Accepted: Nov. 23, 2014 Published: Dec. 1, 2014
Abstract- The cross sensitivity of temperature and ionic concentration are problematic issues in
admittance type level measurement. In this paper cross sensitivity of temperature and ionic
concentration in admittance type level measurement has been thoroughly studied. From the analysis of
results a methodology for elimination of cross sensitivity has developed. This proposes fuzzy based
lineariser to eliminate errors due to temperature and conductivity in admittance type level
measurement. The developed fuzzy based lineariser found an effective tool which is knowledge based
and significantly eliminates high cross sensitivity of temperature and ionic concentration error in
admittance type level sensor. Single electrodes and double electrodes level transmitter cases are also
studied. The result achieved is satisfactory and will help to design suitable smart admittance type level
measurement system for industrial processes.
Keywords: Admittance, level transmitter, fuzzy logic, linearization, cross sensitivity, boiler drum.
INTERNATIONAL JOURNAL ON SMART SENSING AND INTELLIGENT SYSTEMS VOL. 7, NO. 4, DECEMBER 2014
2035
I. INTRODUCTION
Boiler drum level is one of the important parameter to be measured and controlled in process
industry like power plant. The boiler is used to convert water into steam, which is used to turn a
steam turbine eventually generating electricity. Water level control is an important parameter for
operation of boiler effectively. An objective for safe and efficient boiler operation is to maintain
a constant level of water in the boiler drum. If the level is too low, boiler tubes will be damaged
by overheating. Low water conditions have been the most common cause of boiler accidents for
decades. Typically the incidents have occurred as a result of insufficient or inoperable controls.
Many times it is a symptom of improper selection of devices that results in poor performance and
a low availability factor for the safety systems. Instances have occurred where as a result of
continued failures, devices are removed from service due to frustration caused by continued poor
performance even after extensive maintenance effort. This study is intended to direct the design
engineer towards the basic concerns and methods related to specifying Boiler Drum level
measurement [1]. There are different techniques to measure the drum level like conductivity
probe type level transmitter, sight gauges, magnetic type level transmitters etc. [2, 3, and 4].
Unfortunately boiler drum level control is complicated by changes in electrical load requirements
or variation in the fuel and air supply. Hydrastep is a very common water level measuring system
for high pressure or high temperature boiler drum has been used in power plants. There is a need
of continuous level monitoring and easier installation in boiler drum level transmitter for safe
and efficient operation of boiler.
Admittance type level transmitter is one of the low cost level measuring units [5].In case of
admittance level transmitter, the admittance value obtained is proportional to the level of boiler
drum. But the electrical parameter “admittance” has cross sensitivity with other suitable
measurement. Cross sensitivity is the effect of one measurand on the other measurand. Therefore
the admittance level transmitter shows cross sensitivity of temperature and ionic concentration in
boiler drum [6, 7]. The advantage of using admittance type level transmitter is primarily
continuous measurement of level, low cost and can be used in both metallic and non-metallic
tank. Earlier workers tried to develop Admittance type continuous level measurement [8] system
in metallic or non-metallic storage vessel, which is one of the low cost sensing methods for
continuous monitoring the level with some significant cross sensitivity of temperature and ionic
concentration.
The present work has been divided into three parts. Primarily a review on cross sensitivity of
temperature, secondly cross sensitivity of ionic concentration and eventually develop a
computational tool to eliminate these errors which is a fuzzy based linearizer.
Joyanta kumar Roy and Bansari Deb Majumder, ELIMINATION OF CROSS SENSITIVITY IN ADMITTANCE TYPE LEVEL MEASUREMENT USING FUZZY BASED LINEARISER
2036
II. MOTIVATION
In power plants, for measuring the boiler drum level Hydrastep type level transmitter is used
which is costly and very complex level measuring system. This motivates us to use some low
cost level measuring system in boiler drum. Previous researchers already suggested admittance
level transmitter as low cost measuring system. So we have tried to use admittance level
measuring system for measuring the level in boiler drum. But cross sensitivity of temperature
and ionic concentration was a critical issue in case of admittance type level transmitter. Earlier
authors have studied [6, 7]this cross sensitivity effect and found a significant effect of change in
temperature and ionic concentration on admittance type level measurement. We have focused
on admittance type level transmitter and try to eliminate all the cross sensitivity effect to
measure the level accurately and precisely. Meanwhile we have also develop a fuzzy based
linearizer which is multi input system can be incorporated online for elimination of errors from
the measurand.
III. CROSS SENSITIVITY OF TEMPERATURE
The study of cross sensitivity of temperature is carried out initially for single electrode type
admittance measurement system where we have considered a cylindrical metallic electrode A of
radius r is immersed to a depth h in a liquid of permittivity and conductivity σ in a cylindrical
metallic storage vessel of diameter D as shown in figure-1
The impedance between the electrode and the vessel is given by
Where, YParallel is the admittance between the electrode and the vessel. Cparallel is the equivalent of
fringe capacitance and double layer capacitance between the electrodes and the vessel. Rparallelis
Figure.1: Single electrode in a conducting vessel [1]
INTERNATIONAL JOURNAL ON SMART SENSING AND INTELLIGENT SYSTEMS VOL. 7, NO. 4, DECEMBER 2014
2037
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10
0.5
1
1.5
2
2.5
3
Height of the liquid level (in metre)
Ad
mit
tan
ce (
in m
ho
)
Relation of Admittance and height of the liquid level
the equivalent resistance of fringe, electrode double layer resistance between the electrode and
the vessel.
The equivalent admittance between the electrode and the vessel is given by
Where,Yparallel is the admittance between the electrode and the vessel,
h is the height of the liquid level
d is the separation between the electrodes
r is the radius of the electrode
σ is the conductivity of the liquid
is the permittivity of the liquid
The magnitude of the admittance parallel is given by
The above equation is rewritten as
Where,
For admittance transmitter, the relation between the admittance and height of the liquid level is
very much linear in nature [5] represented by the Fig 1.
Figure.2: Plot between Admittance and height of the liquid level for single electrode
of admittance type level measurement with change at fixed ionic concentration and
temperature [5]
Joyanta kumar Roy and Bansari Deb Majumder, ELIMINATION OF CROSS SENSITIVITY IN ADMITTANCE TYPE LEVEL MEASUREMENT USING FUZZY BASED LINEARISER
2038
Admittance depends on conductivity and permittivity of the ionic solution [9][10]. The
conductivity of the solution is related to temperature shown in equation (5).
σTref = * σT(5)
where σTref is conductivity at reference temperature, is Temperature coefficient.
The variation of permittivity with temperature is shown in equation (6).
( T)1/2= 8.851*d(T) (6)
Where, [d (T)]2= 1-0.4536.10-2(T-25)+ 0.9319.10-7(T-25)2
Substituting the variation of conductivity from equation (5) and variation of permittivity from
equation (6) in equation (4),
σTref)2 + 2{(8.851)2d(T)2} (7)
Simplifying the above equation and shown in equation (8).
(T) 2
+2
(T) 2
(8)
Computing the above equation in Matlab, considering different temperature ranges from 30
degree Celsius to 60 degree Celsius. The figure 2 shows the error curve obtained with significant
temperature effect on admittance measurement. The error in Admittance measurement due to the
effect of temperature has been calculated by taking difference of equation-8 and equation -4 and
computing the error curve using MATLAB for single electrode measurement system. Figure 3
shows the effect of cross sensitivity of temperature on admittance level measurement.
It is found from the study that the error of the single electrode admittance type level sensor is
with slope 0.5 and the error of the double electrode admittance type level sensor is (-20% to
42%) with slope 0.67. Therefore It can be concluded that the cross sensitivity of the single
electrode admittance type level measurement is less with double electrode type. The study
reveals that the admittance type level measurement has sufficient cross sensitivity of
temperature. This computational analysis is carried in Matlab version R2010.
INTERNATIONAL JOURNAL ON SMART SENSING AND INTELLIGENT SYSTEMS VOL. 7, NO. 4, DECEMBER 2014
2039
Figure.4: Plot between Admittance of single electrode type and height of
the liquid level with different values of temperature from 30 degree C to
60 degree C [6]
Figure 3: Error curve for temperature effect on admittance level
measurement [6]
Joyanta kumar Roy and Bansari Deb Majumder, ELIMINATION OF CROSS SENSITIVITY IN ADMITTANCE TYPE LEVEL MEASUREMENT USING FUZZY BASED LINEARISER
2040
IV. CROSS SENSITIVITY OF IONIC CONCENTRATION
Similar analysis of cross sensitivity is carried out for admittance measurement having cross
sensitivity of ionic concentration. We have considered same metallic storage tank containing
single electrode, where admittance value is shown in equation (4).
Admittance depends on conductivity and permittivity of the liquid. It is noted that if there is no
change in temperature and ionic concentration then admittance within the two electrodes is
proportional to level of the tank. Here the variation of ionic concentration on the conductivity
and permittivity of the liquid is considered at a constant temperature.
Relation between conductivity of the liquid and the variation of ionic concentration [11] in the
solution is given by equation (9).
C=A0+A1 (σref) +A2 (σref)2 (9)
Where C = Ionic concentration
σref= conductivity corrected to Tref temperature
A0, A1and A2 are specific constants depend on the type of solvents present in the
solution.
There are two roots of the quadratic equation and variation is shown in equation [5].
In the equation [9] for the physical measurement, the negative roots are not considered because
they give imaginary solution.
Relation between permittivity of the liquid and the variation of ionic concentration [12][13] in
the solution is given by
Where
is the dielectric constant of the liquid due to presence of all dissolved solute in solvents.
₀ is the dielectric constant results from the presence of all neutral species including
solvents and ions pairs.
Ii is the ionic strength on the mole fraction Basis.
Ai and Bi are temperature dependent parameters
xi is the mole fraction of the component i.
The temperature dependent parameters Ai and Bi are related by given equation-9,
INTERNATIONAL JOURNAL ON SMART SENSING AND INTELLIGENT SYSTEMS VOL. 7, NO. 4, DECEMBER 2014
2041
Where T is temperature in Kelvin
ai1, ai2, bi1 and bi2 are the parameters determined from experimental dielectric constant data for
electrolytic solutions. By analyzing experimental data that constant values of the parameters can
be used for any of the ionic species in all of the aqueous solutions.
The Yparallel is represented as
Where, σc is the molar conductivity of solution considering the effect of change in ionic
concentration. c is the permittivity of the solution considering the effect of change in ionic
concentration.
The cross sensitivity of ionic concentration have significant effect too on the admittance level
transmitter. For fixed values of liquid level, % error in admittance can be computed by taking the
difference of the admittance ideal (without the variation of ionic concentration) and actual
admittance (with the effect of variation of ionic concentration). The difference value is the
divided by the ideal admittance value to calculate the error. The figure 5 shows the error curve
obtained by varying the ionic concentration in Nacl solution for admittance level measurement.
Figure 6 shows the effect of cross sensitivity of ionic concentration on admittance level
measurement.
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2-40
-20
0
20
40
60
80
100
TDS ( Ionic concentration)
%E
rro
r o
f ad
mit
tan
ce
Error curve of admittance due to the variation of ionic concentration in Nacl solution
Figure.5: Error curve of admittance of the liquid at different range of
total dissolved solids in the solution of NaCl [7]
Joyanta kumar Roy and Bansari Deb Majumder, ELIMINATION OF CROSS SENSITIVITY IN ADMITTANCE TYPE LEVEL MEASUREMENT USING FUZZY BASED LINEARISER
2042
This study reveals that there is significant cross sensitivity after certain ionic concentration
similar cross sensitivity like temperature. For Boiler drum level measurement and control
correction of
Temperature cross sensitivity is required but for other application using normal drinking or non-
drinkable water correction of ionic concentration and temperature are required. The study is
computational based done in MATLAB version R2010.
Our attempt is to eliminate these cross sensitivity effect and suggest some computational tool to
eliminate the errors. For this we have chosen fussy based lineariser which is an effective tool for
error elimination.
V. INTRODUCTION TO FUZZY BASED LINEARIZATION
Linearization is the method of error removal from the measured value of the process
parameter. Fuzzy logic deals with knowledge based computation technique[13] [14] can be
suitable for linearization, error analysis and elimination. Fuzzy lineariser includes input
variables, rule base or inference engine and output variables. There exist knowledge based
systems which related the nature of input and output variables at different condition. The fuzzy
based lineariser designed and computed in MATLAB version 7.10.0 (R2010a).
Here we have considered three inputs variables “A “represents ideal admittance variable ,
“Aerror(T) “ represents measured admittance error of change in temperature and “Aerror(I)”
0 0.5 1 1.5 20
1
2
3
4
5
6
7
8
Height of the liquid level (in metre)
Ad
mit
tan
ce (
in m
ho
)
Liquid level versus admittance at different ionic molar concentration
c=2
c=0.1
Figure.6: Plot between Admittance of single electrode type and height of the
liquid level with different values of molar concentration c=0.1 to 2[7]
INTERNATIONAL JOURNAL ON SMART SENSING AND INTELLIGENT SYSTEMS VOL. 7, NO. 4, DECEMBER 2014
2043
represents measured admittance error of change in ionic concentration. The input variables are
feed to the fuzzy rule base(mamdani) named “admittancenew” .there is one output variable
named”Aoutput” represents the fuzzy corrected admittance value. The MATLAB view of the
fuzzy linearizer(.fis file) is shown in figure 6.
Now the membership functions of each of the input and output variables are framed as shown in
figure 7. In this figure we have selected Aerror(I) input variable and considered five membership
functions whose ranges are considered from the error curve of Admittance with varying ionic
concentration i.e., figure 5.
Figure.6: The fuzzy based linearizer by MATLAB
Figure.7: The membership functions of input and output variables of fuzzy based
linearizer
Joyanta kumar Roy and Bansari Deb Majumder, ELIMINATION OF CROSS SENSITIVITY IN ADMITTANCE TYPE LEVEL MEASUREMENT USING FUZZY BASED LINEARISER
2044
Similarly other input and output variables “Aerror(T)”,”A” and “Aoutput” are represented by the
membership function designed on the basis of the characteristic curve obtained for each of the
parameter. The rule matrix for the input and output variables have been prepared as per
convention that when any of the error is positive in nature Aoutput is subtracted by error value
i.e., Aoutput is decreased. Similarly when any of the error is negative in nature Aoutput is added
by error value and result is obtained or we can say Aoutput is increased. By using these logics
around 56 rules have been prepared and shown in figure 8.
VI. RESULT ANALYSIS
The surface plot of Yerror(I) i.e., Admittance error due to varying TDS of the liquid is varied with
measured admittance (Y) and Youtput i.e., fuzzy corrected data is shown in figure 9 and the
surface plot showing Yerror(T) i.e., Admittance error due to varying temperature the liquid is
varied with measured admittance (Y) and Youtput i.e., fuzzy corrected data is shown in figure 10.
Figure 8: The Rule viewer page of Matlab.
INTERNATIONAL JOURNAL ON SMART SENSING AND INTELLIGENT SYSTEMS VOL. 7, NO. 4, DECEMBER 2014
2045
Now, data’s are recorded for the values of the Admittance value from the lineariser output i.e., Y
corrected from fuzzy lineariser from the rule viewer window, Y ideal from the values evaluated
from the equation shows the relation between admittance and height of the liquid level in meter
and measured admittance(Y) with varying ionic concentration from the equation of admittance
Figure 9: The surface plot of Aerror(I) and A with relation with Aoutput
Figure 10: The surface plot of Aerror(T) and A with relation with Aoutput
Joyanta kumar Roy and Bansari Deb Majumder, ELIMINATION OF CROSS SENSITIVITY IN ADMITTANCE TYPE LEVEL MEASUREMENT USING FUZZY BASED LINEARISER
2046
and height of the liquid level in meter considering only the change of ionic concentration at
constant temperature. The recorded data’s are plotted in figure 11 and figure 12 at varying ionic
concentration and temperature respectively.
The figure 11 shows the comparison study between ideal admittance data with actual admittance
data and also with fuzzy corrected data keeping the temperature constant at 21.9deg.C. Here the
value of ionic concentration (TDS) is varied though out the recorded data. Statistical analysis of
the recorded data have been done and found a Standard deviation of 0.00769 which is under
tolerance range. Now the same analysis have been carried out with fuzzy linearized data at
constant ionic concentration (TDS) and shown graphically in figure 12.
0.0000
0.786(TDS)
0.806(TDS)
0.821(TDS)
1.26
1.36 (TDS)
1.42(TDS)
1.82
2.1(TDS)
)
0.0000
0.5000
1.0000
1.5000
2.0000
2.5000
3.0000
0.00 0.20 0.40 0.60 0.80 1.00
Leve
l se
nso
r A
dn
itta
nce
(Y
)
Liquid level (m)
Comparison study of ideal-vs- actual-vs- fuzzy corrected data
Yparallel without error(Ideal)
Measured (Y) withvarrying ionicconcentration
**All the datas are collected keeping the temperature constant=21.9 deg.C Min error=-0.075 Max error=0.0459 Std Deviation=0.0769
Figure 11: Comparison study of ideal-vs.-actual-vs. fuzzy corrected data at varying
ionic concentration (TDS)
INTERNATIONAL JOURNAL ON SMART SENSING AND INTELLIGENT SYSTEMS VOL. 7, NO. 4, DECEMBER 2014
2047
In the figure 12, the statistical analysis have been carried out and found a standard deviation of
0.0904.
VII. CONCLUSION
The target of the study is to find out nature of errors in admittance type measurement of level due
to cross sensitivity of temperature and ionic concentration, theoretical analysis of the study
reveals the nature of such variation due to cross sensitivity demands practical validation though
experiments. It is observed that fuzzy based linearization technique developed has high
linearizing ability and eliminated the error of temperature and ionic concentration from the
measured value more accurately. The standard deviations noted are within the tolerance range.
The next target of the study will be implementation of theatrical knowledge in developing
experimental setup in suitable laboratory scale for measurement of process liquid level with
variation of temperature and conductivity. The admittance and other parameters will be
converted to digital data, electronic data acquisition system and fuzzy linearization will be taken
in action using suitable data acquisition and graphical analysis software platform. Elimination of
the error of temperature and ionic concentration is one of the requirements for proper
measurement of level in boiler drum. These effects of temperature and ionic concentration also
20.7˚C
25.1˚C
34.8˚C 39˚C
50˚C
53.7˚C
55.2˚C
57.3˚C
y = 3.2888x - 0.0971 R² = 0.9845
0
0.5
1
1.5
2
2.5
3
0 0.2 0.4 0.6 0.8 1
Ideal admittance(inmho)MeasuredAdmittance(in mho)CorrectedAdmittance(in mho)Linear (CorrectedAdmittance(in mho))Linear (CorrectedAdmittance(in mho))
Comparitive study of Ideal Admittance-vs-Measured Admittance -vs-Corrected Admittance (fuzzy linearised data)
Liquid Level (in metre)
Ad
mit
tan
ce (
in m
ho
)
**All the datas are collected keeping the ionic concentration at 0.184TDS Min error=-0.2337 Max error=0.0673 Std Deviation=0.090419869
Figure 12: Comparison study of ideal-vs.-actual-vs.- fuzzy corrected data at varying
temperature
Joyanta kumar Roy and Bansari Deb Majumder, ELIMINATION OF CROSS SENSITIVITY IN ADMITTANCE TYPE LEVEL MEASUREMENT USING FUZZY BASED LINEARISER
2048
give an advantage to measure level, temperature and quality of the boiler water simultaneously.
This method of sensing three or more parameter called as multifunction sensor [16] which is one
of the significant areas to work. There are many applications of multifunction sensors in today’s
world and hence is the future scope of the work.
Finally the target is to develop low cost but continuous level transmitter suitable for high steam
temperature and pressure.
REFERENCE
[1] The Clark-Reliance Corporation, “A guide to boiler drums level equipment and control
concepts”, The boiler drum level measurement guide book, 2001.
[2] Wojciech Skierucha, “Time Domain Reflectometry: temperature-dependent Measurements
of Soil Dielectric Permittivity”, Electromagnetic waves, Institute of Agrophysics, Polish
Academy of Sciences, Poland, Chapter 17, pp374-379,2011.
[3] Lu Guirong, ZhengXianshan, “A new proposal for monitoring oil-temperature and oil-
level”, ISBN:978-1-4244-8158-3,ICEMI-2011,pp 350-353.
[4] Satish Chandra Bera, Jayanta Kumar Ray, Subrata Chattopadhyay, “A low-cost noncontact
capacitance-type level transducer for a conducting liquid”, Instrumentation and
Measurement, IEEE Transactions, volume55, Issue 3, pp-778-786, 2006
[5] Joyanta kumar Roy, “Low Cost Sensing Techniques of Industrial Process Variables”, ISBN:
978-3-659-11192-1LAMBERT Academic Publishing, TP-395, 2012
[6] Joyanta Kumar Roy and Bansari Deb, “Investigation of cross sensitivity of single and
double electrode of admittance type level measurement”, Sixth International conference on
sensing technology , Kolkata, India, Dec 2012, Proceedings published in IEEE Digital
Xplore, ISBN: 978-1-4673-22454, pp- 234 – 237.
[7] Joyanta Kumar Roy and Bansari Deb Majumder, “Cross sensitivity of Ionic concentration
on Admittance type level measurement”, Eighth International conference on sensing
technology ,Liverpool, UK September 2014. Proceedings published in International Journal
on Smart and Intelligent Systems, ISSN: 1178-5608, pp-41-45,
[8] Bera, S. C & Roy. J. K, “Study of an Admittance type Single Electrode Transducer for
continuous monitoring of liquid level in a metallic storage tank, J of Instn. of Engineers (I) ,
vol.-83, pp 56-60, Jan 2003.
[9] Conductivity Theory and Practice, Technical Manual, Radiometer Analytical SAS, France,
Tp: 5-36.
INTERNATIONAL JOURNAL ON SMART SENSING AND INTELLIGENT SYSTEMS VOL. 7, NO. 4, DECEMBER 2014
2049
[10] Matthias Weiß and Reinhard Knöchel, A Novel Method of Determining the Permittivity of
Liquids, IEEE transaction of measurement and instrumentation, Vol. 49, pp-488-492, No. 3,
June- 2000.
[11]Patricia Dalmas, “Conductivity measurement on pure water according to the
recommendations of the USP Pharmacopoeia USP24-NF19”, Article prepared in
international laboratory news , August 2000 edition.
[12] Peiming Wang, AndrzejAnderko, “Computation of dielectric constants of solvent
mixtures and electrolyte solutions”, Fluid Phase Equilibria, ELSEVIER, 186, pp103-
122,2001.
[13] M. Aghaie, S. Ghafoorian, B. Sh. Broojeni and H. Aghaie, “ The effect of dielectric
constant and ionic strength on the solubility of lithium carbonate at 25.00C in
thermodynamic view”, Journal of physical and theoretical chemistry, winter 2009.
[14] Zedeh, L.A, “Knowledge representation in fuzzy logic”, IEEE Transactions on knowledge
and data engineering, volume 1, Issue 1, pp-89-100, Mar 1989.
[15] Carlsson, C, “Fuzzy logic and hyper knowledge: a new, effective paradigm for active
DSS”, IEEE environmental management, 1997, volume 5, pp-324-333, print ISBN: 0-8186-
7743-0.
[16] M.Iwahara, S.C.Mukhopadhyay, S.Yamada and F.P.Dawson, "Development of Passive
Fault Current Limiter in Parallel Biasing Mode", IEEE Transactions on Magnetics, Vol. 35,
No. 5, pp 3523-3525, September 1999.
[17] Lu, G. “A new proposal of multi-functional level meter”, ISBN:0-7803-7987-X,IEEE
International conference on multisensory fusion and integration for intelligent systems,
2003,pp 209-212.
Joyanta kumar Roy and Bansari Deb Majumder, ELIMINATION OF CROSS SENSITIVITY IN ADMITTANCE TYPE LEVEL MEASUREMENT USING FUZZY BASED LINEARISER
2050