Sei sulla pagina 1di 6

FlowControl Network Page 1 of 6

Free Product Information


Click here to use our NEW online site-wide search
Reader Service and request product
information directly from our featured vendors!

Wednesday

FREE SUBSCRIPTION
ARTICLE ARCHIVES
CURRENT CONTENT
NEXT ISSUE
SUBMISSIONS ARTICLE ARCHIVES Home
ADVISORY BOARD
CONTACT STAFF
PAST ISSUES
EDITORIAL CALENDAR Why Measuring Flow is a Difficult Task M
ADVERTISING INDEX
Run New Search

by: David W. Spitzer, P.E., and Walt Boyes


WEFTEC Pages: 44-47; August, 2002
September September 28 -
October 2, 2002 •A
Chicago, IL It is a widely held perception that sensors producing traditional
ISA •C
October 21-24 measurements are easily selected, purchased, installed and operated. •F
Chicago, IL This perception seems to drive the current trend of shifting this work
Search for More Events •F
from instrumentation engineers to technicians who often have little or no
•G
supervision, and sometimes little or no training in understanding
sensors and measurements. •L
•M
Logically, there is some truth to the argument of shifting the •P
work to technicians. After all, anybody can put in a pressure gauge -- •P
right? Perhaps in most cases, but what about the application where an •P
important pressure gauge was checked by technicians every month and • R
replaced every one to two months? Installation of a remote pressure F
gauge with proper materials of construction not only paid for itself in •S
about six months, but also increased operator safety. Similar examples •S
can be produced for any of the four most common measurement •S
variables: flow, level, pressure and temperature. Each of these •S
measurement variables has its idiosyncrasies. •T
•T
Despite level, pressure and temperature measurements being •V
more prevalent, the measurement of flow often presents the greatest
challenges -- and can result in some of the greatest rewards.

Measuring flow is not easy. The barriers to achieving accurate


and reliable flow measurements are many and varied, and even worse,
these problems and issues are often interrelated.

Steady-state flow is a dynamic process variable. Steady-state


flow describes the dynamic movement of fluid, where, in contrast,
steady-state level, pressure and temperature fundamentally describe
conditions that are static.

http://www.flowcontrolnetwork.com/article_index.cfm?article=475 7/31/02
FlowControl Network Page 2 of 6

By their nature, steady-state static process variable conditions


tend to be relatively free of noise. An example of this is a water level
measurement without turbulence or other “noise” producers. In contrast,
because of the dynamic nature of steady-state flow, the existence of
“flow noise” in the flow measurement signal is such a common
phenomenon that we often do not recognize it as “noise.” But when
similar noise occurs in other process variable measurements, it is
unusual enough that the signal is quickly termed “noisy.”

Damping the Signal

In either case, various techniques can be used to reduce the


noise component of the signal. One method is to dampen the signal.
This method will usually work in most applications, because the noise
will be reduced as the damping is increased. Also reduced is the ability
of the instrument to measure the process in a timely manner because
the response time of the damped instrument will increase. Typically,
when more than a small amount of damping is necessary to reduce
noise, there is another problem. In these applications, increasing
damping can mask the real problem and could adversely affect the
process.

In flow measurement, damping is usually required to achieve a


usable signal for a controller. The user must then decide where the
damping will be installed -- in the control system, in the flowmeter or
both. Damping performed in the control system is usually readily
accessible, and its values can be documented as part of the control
system documentation. However, the damping algorithms in the
flowmeter are likely to be superior to the generic algorithm(s) found in
the control system.

When damping is installed in the flowmeter, the damping value


is typically not readily accessible. The damping value(s) should be
appropriately documented so that the information is available when the
flowmeter is repaired or replaced. Further, when large amounts of
damping are added in the flowmeter, changes to the flowmeter and/or
process that cause the flowmeter itself to become unresponsive may be
masked. Thus, there appears to be no one best location to implement
damping.

Another alternative is to implement damping in the flowmeter


and in the control system. The amount of damping installed in the
flowmeter should be relatively small to attenuate the higher frequency
flow noise. The signal should be of sufficient quality to use for recording
purposes and might still “paint” a small band on a chart. A small amount
of additional damping may be added in the control system. An
advantage of locating some damping in both locations is that the flow
signal is responsive (but not overly so) and can be observed (and
adjusted) in the control system. Disadvantages include additional
adjustment and documentation associated with the two damping
locations.

http://www.flowcontrolnetwork.com/article_index.cfm?article=475 7/31/02
FlowControl Network Page 3 of 6

In control loops, flow noise can cause unwanted control valve


movement because the controller flow noise contributes directly to the
controller error calculation. Such error affects the controller output. An
almost universal technique to reduce the effect of flow noise in flow
control loops is to eliminate the use of the derivative (rate) in flow
controllers. The derivative (rate) adjusts the control valve based upon
the rate of change of the error. Because the process variable directly
affects the error, flow noise (that generally changes relatively rapidly)
can cause the controller output to change rapidly. In addition to process
problems, this can result in rapid valve movement and excessive valve
wear.

Profile Effects

Pipe hydraulic problems are among the leading causes of


flowmeter problems. Flowmeters are designed to operate in piping
where the flow entering the flowmeter exhibits a homogeneous and
symmetrical velocity profile.

While there are some flowmeters that are immune to effects of


velocity profile, the performance of many flowmeters can and are
significantly affected by distortion of the velocity profile that enters the
flowmeter. To ensure an adequate velocity profile, many flowmeters
require straight run upstream and downstream of the flowmeter. The
rule of thumb that flowmeters need 10 diameters upstream and 5
diameters downstream is a myth. Some flowmeters may need upwards
of 40 diameters upstream to ensure that the velocity profile entering the
flowmeter is adequate. Other flowmeters may need less than 5
diameters. These values are themselves not correct in all locations.
Studies have shown velocity profile distortion caused by piping
restrictions located over 100 diameters upstream.

Control valves and fittings create turbulence that distorts the


velocity profile in the pipe. They should be located downstream of the
flowmeter. For similar reasons, thermowells are generally located
downstream of the flowmeter, even though a slight temperature
measurement error can occur.

Physical properties can render almost any measurement


instrument inoperable. Problems can occur when the physical
properties of the fluid are outside the limits of the flowmeter. Applicable
physical properties include pressure, temperature, viscosity, specific
gravity, lubricity, thermal conductivity, electrical conductivity and the
like. Only one constraint need be exceeded to cause excessive
measurement error and/or damage the flowmeter. For example, even
momentarily operating a plastic flowmeter at 500°C will likely damage or
destroy the flowmeter.

Process issues can have a major effect on flowmeter


performance. Fluid composition can affect the physical properties of the
fluid and adversely affect the measurement. Flow rate can affect the
hydraulic characteristics of the fluid in the pipe, and cause the velocity

http://www.flowcontrolnetwork.com/article_index.cfm?article=475 7/31/02
FlowControl Network Page 4 of 6

profile to be different from that in which the flowmeter was designed to


operate. In addition, the state of the fluid can affect the measurement.
For example, consider a flowmeter installed in water service where the
sensing lines freeze. Filling the flowmeter with ice, and not the pressure,
temperature or composition, renders the flowmeter inoperable.

Remember, performance can be degraded and/or the flowmeter


can be damaged if any one of the many limits is exceeded.

Ten Flowmeter Technologies

There are over 10 broad flowmeter technologies from which to


select a flowmeter for a given application. Within each broad technology
are a number of related technologies, each of which are manufactured
by a number of companies. If one assumes that each of 10 broad
technologies contains five related technologies, each of which has five
suppliers, there are 250 possible flowmeters from which to select.
Recent flowmeter studies performed by the authors indicate that there
are over 20 suppliers of Coriolis mass flowmeters, over 40 suppliers of
vortex shedding and other fluidic flowmeters and over 60 suppliers of
magnetic flowmeters. These three technologies alone would account for
about half of the estimated 250 suppliers. The implication is that there
are many more flowmeter suppliers. Further, when the individual
flowmeter models are considered, the number of potential flowmeters
from which to select is well into the thousands.

Notwithstanding the large number of available flowmeters, even


selecting from among the broad flowmeter technologies can be a
challenge. Different technologies react to different operating conditions
and flows differently. Determining which flowmeter technology is most
immune to the actual process conditions in a given application can be
overwhelming, and is often beyond the skill of a technician.

Calibration Complexity

Calibrating flowmeters is complicated by the dynamic nature of


steady-state flow. In comparison, consider that a level transmitter can
typically be calibrated by adjusting the transmitter with known levels in
the tank. Pressure transmitters can be calibrated in the field by
developing a known pressure using a dead-weight tester that distributes
a weight over an area. Temperature transmitters can be calibrated in
the field at 0°C and 100°C by using ice water and boiling water
respectively. These statements may seem somewhat simplistic because
there is quite a bit of detail involved to perform a proper calibration, but
the basic concept is sound.

Flowmeter calibration is much more involved. It is not possible to


take a liter per minute (lpm) into the field and throw it through a
flowmeter to check performance. However, some flowmeters can be
removed and brought into a flow calibration laboratory for calibration. As

http://www.flowcontrolnetwork.com/article_index.cfm?article=475 7/31/02
FlowControl Network Page 5 of 6

compared with the apparatus used to calibrate the other process


variables mentioned, these laboratories are generally more complex
because the weight, density, temperature and composition of the
laboratory fluid and process fluid (and other items) should be taken into
consideration. Field calibration can also be performed using provers
when the piping is appropriately designed.

Due to expense and complexity, flow laboratories and provers


are typically only available when flow measurements are extremely
important, such as in custody transfer applications. Therefore, most
industrial facilities have no way to wet calibrate flowmeters. As a
surrogate, the technician checks for proper calibration based upon
differential pressures, millivolts, frequencies, weights, volumes and the
like. These procedures will not detect a plugged impulse tube, a
deformed orifice plate, a damaged magnetic flowmeter liner and other
potentially serious problems. As such, these commonly performed
transmitter “calibrations” are better than doing nothing, but they do not
check the calibration of the flowmeter. Calibration can be performed by
flowing a known amount of fluid through the flowmeter under test, and
comparing this amount with the amount of flow measured by the
flowmeter.

As previously discussed, some flowmeters require straight run,


while others do not. Installation of control valves, thermowells and the
like upstream of the flowmeter can cause hydraulic turbulence that can
adversely affect many flowmeters.

While flowmeters within some technologies are installed


similarly, many flowmeters have idiosyncrasies that apply only to a
specific flowmeter model. For example, mounting requirements for
Coriolis flowmeters vary widely -- some manufacturers recommend four
supports, while others do not recommend any supports.

With the notable exception of custody transfer installations,


many flowmeters are not installed per manufacturer installation
recommendations. These flowmeters will not necessarily achieve the
level of performance claimed by the manufacturer. One of the most
common causes of installation failure in flow measurement is not paying
attention to the “details.”

Despite performance that may be significantly degraded, users


typically believe that the flowmeter performs within its claimed
performance. When important process diagnosis and operating
adjustments are based on these measurements, the quality of these
measurements could cause inappropriate results that could be
detrimental to the process. For example, in one application, an
improperly installed flowmeter performed over 50 times worse than its
claimed accuracy. Had the quantity of the fluid not been clearly visible,
the user would not have known about the error, nor would the user have
suspected that an error existed.

Many have the mistaken impression that flowmeter selection,


installation and operation is relatively easy. However, even individuals
experienced in flow measurement often have reservations about the

http://www.flowcontrolnetwork.com/article_index.cfm?article=475 7/31/02
FlowControl Network Page 6 of 6

flowmeters they select for a particular application. Achieving the


promise of superior flow measurement is not an accident -- it is paying
attention to the details.

References

1. Flow Measurement (2nd Edition), David W. Spitzer, ISA, 2001


(838 pages).

2. Industrial Flow Measurement, David W. Spitzer, ISA, 1990


(441 pages).

About the Authors

David W. Spitzer, P.E., and Walt Boyes have more than 50


years (combined) experience in instrumentation and process control,
and are the principals of Spitzer and Boyes, offering product
development, marketing, specifying, installing, start-up and
troubleshooting services for manufacturing and automation companies.
Both Mr. Spitzer and Mr. Boyes have written numerous books and
articles, have taught seminars and are members of standardization
committees. Mr. Spitzer can be reached at 845 623-1830 or
dwspitzer@aol.com, while Mr. Boyes can be reached at 425 432-8262
or walt@waltboyes.com. Also, visit www.spitzerandboyes.com
Subscribe | Email | Site Map | Search | FAQ | Advertise

Copyright ©2002, Witter Publishing Corporation


Email: FlowControl@WitterPublishing.com
Flow Control Network
is the technology
resource for the fluid
handling industry's
critical disciplines of
control, containment F
and measurement. wi

http://www.flowcontrolnetwork.com/article_index.cfm?article=475 7/31/02

Potrebbero piacerti anche