You are on page 1of 7

1

FLOW MEASUREMENT 2001 INTERNATIONAL CONFERENCE


ADVANCED DIAGNOSTICS IN MAGMETERS
Thomas Budmiger, Endress+Hauser Flowtec AG, CH-4153 Reinach BL
1 GENERAL ASPECTS
Todays increasing competition and cost pressure in the process industry lead to an concomitant
increase in overall efficiency and more cost-efficient production, too. To achieve this, production
costs will have to be reduced by striving for as high a production rate and as low a percentage of
downtimes and malfunctions as possible while keeping overheads as low as possible.
As a result, the overall costs of a measuring point will have to be considered and optimised over the
entire life cycle of the measuring device, from the selection of the actual device and the project
phase to the necessary maintenance. The purchase price alone is no longer decisive.
Independent studies show that purchase and initial operations only amount to about 30% of the
overall costs over the entire life cycle of a measuring point, with 70% of the overall costs being
caused once the device is installed.
A lot of quite unnecessary maintenance work is done in the name of prevention due to a pre-
planned plant maintenance schedule, although the device shows no malfunction and need not have
been checked. These periodically executed checks cause high fixed costs. Each measuring-point
function and each additional measured value that might conceivably serve to reduce such fixed
costs and the probability of downtimes and malfunctions will help increase the value of the
production plant.
2 INTRODUCTION
A number of measurable parameters are available for advanced diagnostics when using
electromagnetic flowmeters, which may be evaluated according to different criteria. On a sensor,
for instance, the temperature of magnetic system coils may be evaluated as a trend regarding fluid
temperature and the thermal long-term load.
A crucial problem realising advanced diagnostics consists in tracing the cause of a change thus
established. After all, a measurable change may very well have several different causes. For a
correct interpretation of the measured results, the process is to be included in the evaluation, too.
A careful interpretation of the measured values may help you realize substantially advanced
monitoring of the sensor. The result is an early recognition of malfunctions and an optimization of
maintenance intervals.
In addition to the actually measured flow, it is possible to get additional information relevant to the
process. Essentially, indirectly available information is explicitly evaluated according to various
criteria.
2
The representation of
a sensor cross-section
in Fig. 1 shows the
sources of the informa-
tion used for advanced
diagnostics. These are
R
COIL
, the resistance of
the coils to generate
the magnetic field, as
well as U
EL1
and U
EL2
,
the potentials between
the measuring elec-
trodes and the refer-
ence potential, as well
as the relaxation
behaviour on the elec-
trodes.
The measured results
below were collected
over a period of sev-
eral months at a pulp
mill in Switzerland.
The measurement
section is only
activated part of the
time and the measured
fluid was defined as
follows:
Fluid Reduced sulphite waste liquor (calcium lignin
sulphonate)
Concentration 50 53%
Traces of Furfural, cymene, methanol, etc.
Gypsum 0.5%
PH value 2.5 4.5%, depending on the product
Temperature 85 95C (summer / winter)
Pressure 8 bar
A standard Endress+Hauser Flowtec AG magmeter, a PROMAG 53P, DN50, was used as a
flowmeter. The measuring tube liner of the sensor is made of PFA and the measuring electrode of
Hastelloy C. The sensor is mounted in a vertical pipe. The data shown in detail below were read
from the flowmeter and saved at 20 minute intervals.
The data collected by the flowmeter (electrode potentials, relaxation time) may be read from the
local indicator or via various field busses (HART, PROFIBUS, FF) and processed.
3 COIL RESISTANCE AND TEMPERATURE
The resistance of the magnetic system coils is highly suitable for monitoring the general thermal
load of the sensor. Note, however, that the temperature calculated on the basis of the coil
resistance is not equal to the fluid temperature. Relevant parameters affecting the temperature
Fig. 1 - measured variables which may be evaluated for
advanced diagnostics
3
include outside warming by the fluid, self-warming of coils by the electrical power applied, and the
ambient temperature. Nevertheless, the general thermal load is an indicator of the life of the
sensor.
Upon comparison with reference data, the data collected by the flowmeter permits conclusions as to
the condition of the sensor and, as far as it concerns the thermal load, its life expectancy. Available
information includes:
1. Minimum coil temperature min (T
COIL
) in C
2. Maximum coil temperature max (T
COIL
) in C
3. TLF (thermal load factor) ( ) t T T
NOMINAL COIL

in Ch
4. TLCF (thermal-load change factor)

t
T
COIL
in C / h
5. Operating hours of the sensor in hours
Limit values 1 and 2 show the
thermal excess load of the
sensor. Since a maximum
admissible fluid temperature
depends on the liner material,
the respective limit values
have to correlate with this
parameter. If the respective
limit value is exceeded, the
lifetime of the sensor will be
reduced. The more frequently
and the longer such a limit
value is exceeded, the likelier
that instrument downtime or
malfunction will occur. The
thermal load factor (TLF)
equals the product of the
measured temperature of the sensor coils and the operating time. It is important information on the
permanent load of the sensor. The thermal load change factor (TLCF) shows the frequency and
speed of changes in temperature.
The mean thermal load of the
sensor over the entire
operating period may be
calculated by dividing the
TLF by the number of
operating hours. Fig. 2
shows a simulation of such a
temperature curve.
The values resulting from the
coil temperature for TLF and
TLCF are shown in Fig. 3. In
case the power should be
interrupted, the last data
collected will be saved in the
sensor memory. In this
manner, the entire life cycle
of a sensor may be recorded in a compressed manner.
0
20
40
60
80
100
120
140
160
0 5 10 15 20 25 30 35 40
hours
C
o
i
l

T
e
m
p
e
r
a
t
u
r
e

d
e
g

C
Fig. 2 - coil temperature
0
20
40
60
80
100
120
140
160
0 5 10 15 20 25 30 35 40
hours
TLF
TLCF
Fig. 3 - cumulated thermal load or change from Fig. 2
4
4 ELECTRODE POTENTIAL IN THE FLUID
By measuring the electrode potentials, sealing problems at the measuring electrodes may be
recognised, and short circuits in the wiring or at the sensor, caused by conductive deposits may be
detected early on. Besides, serious changes in fluid properties may lead to significant changes of
the electrochemical electrode potential. Such changes may be detected and evaluated by
continually monitoring the potentials.
The potentials between the measuring electrodes and the reference potential are composed of the
electrochemical voltage of the galvanic elements electrode/fluid, the measuring signal, as well as
additional interference signals of varied provenance (resistive via the fluid, inductive and/or
capacitively coupled). In particular, the lowest frequency parts are highly dependent on the fluid and
the material of the measuring electrodes.
The relevant voltage difference between the two measuring electrodes to determine fluid velocity is
very low compared to the mentioned interference signals (typically ! 300V / ms
-1
) and, as a rule,
registered by integration over a fixed period of time. This efficiently suppresses higher-frequency
interference. Lower-frequency interference signals are eliminated by establishing the difference of
two succeeding measuring periods with opposite orientation of the magnetic field. The absolute
potentials on the measuring electrodes are not included.
Recording the individual electrode potentials poses no great problem. The processing effort and
the interpretation of data depends on the resolution and the rate of digitisation. Working with a
relatively low resolution and a low rate of digitalization, only slow processes may be observed and
interpreted accordingly.
Fig. 5 ff. shows the
measured potential on
the measuring elec-
trodes. The measure-
ment section was
emptied at irregular
intervals and the pipe
rinsed with hot water
while the measure-
ment section was
bypassed. In addition,
longer periods of up to
4 days occurred with
empty measuring
tubes (see Figs. 4 and
5) with a range of
460 560 h. The
measured voltage of
1.6 volt equals an
empty pipe.
-1
-0.5
0
0.5
1
1.5
2
2.5
3
3.5
4
0 200 400 600 800 1000 1200
hours
V
o
l
u
m
e

F
l
o
w

(
m
/
s
)
Fig. 4 - measured flow in the test phase
5
The sequence of poten-
tials on the measuring
electrodes during active
operations may be classi-
fied into 2 different types.
One type shows only few
differences of potentials
which slowly rise com-
pared to the reference
potential of ! 1 mV / h, up
to an absolute value of
<100 mV (also see detail
of Fig. 5 in Fig. 6).
The second type
shows significantly
different behaviour.
The voltage of the two
measuring electrodes
compared to the refer-
ence potential shows a
difference of 100 mV.
The absolute value is
in a relatively wide
range between
-100 mV500 mV
(see Figs. 5 and 7).
The reasons for this
divergent behaviour is
shown in Fig. 4. One
type (Fig. 6) occurs
when fluid flows, the
second when it stands
still (Fig. 7). Small
differences in the area
of the input circuits of
the two measuring
electrodes cause an
asymmetric load of the
interface capacities. It is
not reduced in case of a
standing fluid and a set
rate of repetition
(1 pulse per hour). If
the fluid however
passes the electrodes at
a flow rate of about 2 ms
-1
, these asymmetries completely disappear.
-2
-1.5
-1
-0.5
0
0.5
1
0 200 400 600 800 1000 1200
hours
v
o
l
t
s
Fig. 5 - potentials between measuring electrodes
E1 or E2 and GND
0
0.02
0.04
0.06
0.08
0.1
0.12
600 620 640 660 680 700 720
hours
v
o
l
t
s
Fig. 6 - 1
st
detail view of Fig. 5
-0.1
-0.05
0
0.05
0.1
0.15
80 100 120 140 160 180 200 220
hours
v
o
l
t
s
Fig. 7 - 2
nd
detail view of Fig. 5
6
The potentials on the measuring electrodes to date recorded by this field test show a rather
inconspicuous behaviour, conforming to the flow data measured. In case of an empty measuring
pipe, the data collected however lose their information value. However, this can be taken into
account by activation of the EPD insofar as exceeding or falling below a limit value allocated to this
measured variable will not activate a report.
5 LOAD RELAXATION IN THE MEDIUM
The detection of electrode coating is handled indirectly by way of the dielectric properties. It is
realised by a controlled change of the potential on the measuring electrodes and by measuring the
succeeding period of relaxation of the potentials. The duration of the potential shift may be set to fit
the fluid. As a rule, this is a matter of some milliseconds. Since the relaxation time largely depends
on the respective process, any detection of deposits has to be set within the process. This is done
by repeatedly applying a test pulse of varied length and by measuring the relaxation time until the
latter is within a pre-defined window. The resulting length of the test pulse as well as the
concomitant relaxation time are saved as reference variables.
Actual monitoring is handled by a regular application of the test pulse at a pre-set length and by
measuring the relaxation time. The relaxation time is then compared to the reference time. If the
time measured in the course of a pre-set period does significantly deviate from the reference time, a
report is made. Due to the low gain of the amplifier or the extremely high dynamic range of the
measuring amplifier (> + 1.5 V on the measuring electrodes), the pulse to measure the relaxation
will not affect the flow measurements.
The measured results of the relaxation time show a substantial dependency on the fluid velocity. At
2 m.s
-1
it is within a range of ! 100 300 seconds, in the case of a standing fluid, the relaxation
time is larger by one order of magnitude. The parameters relevant for the tests were set as follows:
Measuring period 60 msec
Polarity of the measuring pulse positive
Duration of the measuring pulse 1.0 msec
Amplitude of the measuring pulse 12 V
Rate of repetition of the relaxation measurement 3600 sec
Discrimination threshold 10%
The results shown in Fig. 8
show a measuring period of
about 50 days. The forma-
tion of a deposit impeding
flow measuring would, at
given process conditions,
cause the trend of the
relaxation time to fall given
conductive deposits
(shorter time) or to rise
given insulating deposits
(longer time).
In the field test shown here,
no essential change of the
relaxation time caused by
such deposits could be
detected to date. The
0.001
0.01
0.1
1
10
100
1000
10000
0 200 400 600 800 1000 1200
hours
s
e
c
Fig. 8 - measured relaxation time of the detection of deposits
on the electrodes
7
results of the flow measurement (see Fig. 4) confirms our assumption that there is no change within
the measuring tube affecting the measurements.
The results of the relaxation measurements show good reproducibility of the measured values, and
frequently occurring
minimum periods of
4 msec. These are
caused by the fluctuations
of the electrochemical
potential that cause the
relaxation measurements
to stop. Fig. 9 clearly
shows that the
measurements stopped
due to a too extensive
variance of the
electrochemical potential
which may amount to an
uninterrupted period of
several hours. Therefore,
continual recording of this
measured value over a
longer period of time will
be necessary. Based on
individual measuring results, a statement as to the formation of deposits in the area of the
measuring electrodes is somewhat uncertain. In addition, test results show that the relaxation time
does not only depend on fluid and electrode properties but also on fluid velocity. The characteristics
of the connection between the relaxation time and the fluid velocity will have to be shown by
additional tests.
6 SUMMARY
The current situation of the advanced diagnostics when using electromagnetic flowmeters shows
clear progress regarding instrument and process monitoring and, thus, additional possibilities of
optimising servicing and maintenance costs.
A collection of coil resistance data will provide information on the temperature on the magnetic coil
and, thus, indirectly on the temperature load of the sensor.
When monitoring the electrode potentials, it is less the absolute value than the trend of the two
potentials which will be of interest. Observing these trends of both electrode potentials might
provide an early indication, along with the knowledge of the respective application, to detect and
avoid damage to the respective electrodes. Thanks to the measured electrode potentials, an
excellent reproducibility and, thus, a basis for an advanced process and system monitoring could be
shown.
First field tests of relaxation behaviour measurement on the measuring electrodes show the
expected dependency of the relaxation time on the respective process conditions. Given constant
process conditions, significant changes in the pulse decay, depending on the process, provide
information on a initial coating formation. However, the know-how of the user is always necessary
to correctly interpret the diagnostic information acquired.
0.001
0.01
0.1
1
10
100
1000
600 620 640 660 680 700 720
hours
s
e
c
Fig. 9 - detail of Fig. 8