GOOD PRACTICE GUIDE AN INTRODUCTION TO FLOW METER INSTALLA INST ALLATION TION EFFE EFFECTS CTS
An Introductory Guide to Flow Meter Installation Eects
The aim o this Good Practice Guide is to introduce the subject o installation eects. Every ow meter is subject to t o some kind o installation eect. A ow meter only stands a chance o living up to the manuacturer’s accuracy and uncertainty claims i it is installed correctly. Installation eects can be large or small but can be dramatically reduced i you give them a little consideration.
Contents Foreword
2
1.
What is an installation eect?
3
2.
Fluid properties
3
3.
How density and viscosity interact with ow
3
4.
Flow regimes
4
5.
Velocity profles
4
6.
Flow development
5
7.
More distorti distortions ons
6
8.
Eects on specifc ow meters
7
8.1
Dierential pressure meters
7
8.2
Electromagnetic ow meters
8
8.3
Ultrasonic meters
8
8.4
Turbine ow meters
8
8.5
Vortex meters
9
8.6
Variable Variab le area meters
9
8.7
Positive displacement meters
9
8.8
Coriolis meters
9
8.9
Thermal mass meters
9.
What can be done about installation eects?
10 10
10. Is there an alternative
10
11. Round up
11
12. Further reading
11
Foreword This is an introducto introductory ry guide or people who know little or nothing about ow meter installation eects, but who need to learn about them. You, You, yoursel may already be all too aware o the importance o installation eects and armed with this document are going to educate others. It is or anyone involved in ow measurement rom engineers, to managers, to sales people, to customers. A lot o time and money can go into the purchase o a ow meter, meter, yet all too oten little consideration is given to its installation. This can lead to the meter over or under reading leading to errors which could could easily have been avoided. In some instances, ow meters have been installed so badly that you would have been as well just using a straight bit o pipe and guessing what the owrate is. To put this into context, imagine you are selling £60 million worth o product a year. A 1% error in your custody transer meter equates to £600,000, a 5% error will cost you £3 millon! That is lost revenue which could have been avoided and is better o in your pocket than someone else’s. This introductory guide guide is by no means the fnal word on installation eects. ar rom it in act. This is merely a starting point to get the reader to start thinking about the importance o installation eects and how this may impact the type o meter chosen. It does not contain all the answers to installation eects and the problems they can cause but does point you in the right direction to where they can be ound. The opening sections o this introductory guide cover uid properties and how they interact in ow. This is because in order to understand installation eects you need to know a little bit about ow. Then we move on to velocity profles, how they develop, and what eects bends and valves have on them. Next, we look at how specifc types o meters and how they are aected by installation, fnishing o with some possible solutions and some sources o urther reading to help you on your way wa y.
Good Practice Guide
1. What is an installation eect? You may have heard the term installation eects with regards to ow measurement but you aren’t sure exactly what this means or why they could be important to you or when you should take them into consideration. This Good Practice Guide (GPG) will remedy that and help you understand what it means and why it is vital that you care. However,, beore trying to understand what an installation eect actually is, it is important to understand a little about uid However properties and ow.
2. Fluid properties
He
The concepts o mass and volume are airly amiliar and encountered in every day lie when buying a pint o beer, flling a car with petrol, or buying a kilogram o apples. Some less well known uid properties are density and viscosity which are important in the understanding o installation eects. Density is the mass o a substance divided by a unit o volume such as litres, gallons, cubic meters, and
Kr
so on. Helium gas is less dense than air hence why when you buy a balloon flled with helium you need to hold on tight or it will oat away. Conversely, krypton is denser than air so i your balloon is flled with krypton it would sink to t o the ground.
= Viscosity is a measure o a uid’s uid’s ability to resist external orces which are trying to change its orm. A simple way to think o viscosity is as “thickness”. Comparing Comparing treacle with water,, we would say that treacle is more viscous than water as it is much water thicker than water. Temperature aects viscosity; heating up treacle will make it thinner or in other words, decrease the viscosity. However, the same does not
LIQUID Temperature
GAS Temperature
Viscosity
Viscosity
apply when it comes to gasses. Increasing the temperature o a gas will actually increase its viscosity.
3. How density and viscosity interact in ow Density relates to the mass o a uid which in turn relates to momentum1 and is what keeps the uid travelling along the pipe. Viscosity on the other hand relates to internal riction which will slow and possibly even stop the uid rom travelling. The interaction between density and viscosity o uids was studied in the 19th Century by Osborne Reynolds, a prominent scientist in the understanding o uid mechanics. He also studied the heat transer between solids and uids but that is not relevant to this particular topic. Reynolds described the relationship between the momentum and retarding orces within a uid and this lead t o the Reynolds number which is the momentum o the uid divided by the viscosity o the uid. The Reynolds number, Re, has no units and is thereore known as a dimensionless number.
An Introductory Guide to Flow Meter Installation Eects
3. How density and viscosity interact in ow (cont.) WARNING! Although Re is dimensionless you need to use consistent units when
or or
2
calculating it. Using SI units throughout helps to keep things simple. Now that we know about the Reynolds number how does it help our
velocity pipe diameter diameter UU= =velocity DD==pipe µ = absolute viscosity = kinematic µ = absolute viscosity viscosity = kinematic viscosity viscosity density = =density
understanding o uid ows?
4. Flow regimes Flow will propagate itsel in one o two ways and these types o propagatio propagation n are reerred to as ow regimes. The Reynolds number can be used to classiy the ow regime. I the Reynolds number number is less than 2000 then the viscous orces are dominant and this means the ow is laminar. Laminar ow can be thought o as the uid moving along in thin layers with no mixing between the layers. It is sometimes reerred to as Poiseulle ow in text books. I the Reynolds number is greater than 5000 then the inertial orces dominate and this means ow is turbulent. With turbulent ow the bulk motion is parallel to the pipe’s axis but with mixing between the layers. This is why turbulent ow is good or heat transer. You may be wondering what happens i your Reynolds number lies between 2,000 and 5,000. This region is known as transitional ow where your ow switches back
Re < 2,000 Laminar
and orth between the laminar and turbulent ow regimes. Transitional ow is less
Re > 5,000 Turbulent
predictable and because o this it is more difcult to meter.
5. Velocity profles A velocity profle describes how ast a uid is owing at dierent points across across the pipe. I there was no riction at all the uid would all move at the same velocity like a solid plug. However, the pipe wall creates riction and a thin layer o uid next to the pipe wall does not move. The urther away rom the pipe wall the uid is, the less riction there is and so the uid moves at increasingly higher higher velocities. This means that the highest velocity occurs at the centre o the pipe i.e. at the urthest point rom the pipe walls. How the velocity profle changes across the pipe depends on the nature o the ow regime; laminar or turbulent.
Good Practice Guide
5. Velocity profles (cont.) As previously discussed, laminar ow occurs when the viscous orces (or “internal riction”) dominate so the velocity o the uid changes gradually rom the pipe wall to the centre line. I you were to plot the velocity profle o a ully developed laminar ow you would see that it is parabolic in shape.
The maximum velocity in the centre o the pipe is twice the average.
However,, with turbulent ow the inertial orces are dominant so the velocity profle is atter in the middle and varies However steeply near the pipe wall.
The maximum velocity in the centre o the pipe is between 1.1 and 1.3 times the average velocity.
6. Flow development A velocity profle takes time to develop but when it does stabilise it is considered to be ully developed which which is also commonly reerred to as an “ideal” profle. An ideal profle develops within a length o pipe which, theoretically, is o infnite length. In reality though, an engineer or designer is looking or the length o pipe which will allow the ow to develop a profle that is very close to ideal. As this length is related to the diameter it is usually expressed as pipe diameter equivalents rather than distance.
D = 0 .2 m
For example, we would say this pipe is 4 diameters long i.e. 4 x 0.2m = 0.8m
0.2m
0.2m
0.2m
0.2m
Pipeline components such as partially closed valves, reducers and bends, distort the velocity profle. Let’s Let’s take an example o a uid travelling around a bend. The uid on the outside o the bend has to move more quickly and this gives a skewed or asymmetric profle.
An Introductory Guide to Flow Meter Installation Eects
6. Flow development (cont.) Most ow meters have been developed to work best under “ideal” conditions. International Standards Organisation (ISO) or the meter manuacturer will advise on the best installation and usually speciy a length o straight pipe upstream o the meter and another length o straight pipe downstream. The specifc ISO standards relating to installation can be ound in the reerence section at the end o this guide. One o the biggest problems with industrial metering systems is a ailure at design stage to appreciate the importance o providing the meter with as near to an ideal ow as possible.
7. More distortions As well as creating an asymmetric velocity profle, ow around a bend can create a secondary ow. The ow coming round the bend is orced against the pipe wall on the outside o the bend and rebounds towards the opposite side o the pipe orming vortices. These departures rom an ideal profle can interact with the ow meter giving signifcant measurement errors.
One bend is bad, two bends are worse!
Two bends in dierent planes cause a distorted profle and a single vortex also known as swirl. Swirl can take well over 100 pipe diameters to decay; in act there have been cases where swirl has persisted or over 500 hundred pipe diameters beore the profle returns to acceptably close to ideal. Swirl tends to aect meters more than asymmetry causing your meter to potentially over or under read signifcantly.. Viscosity o the uid oten has a bearing on the rate o decay o swirl. signifcantly Thin uids, or very low viscosity, viscosity, tend to propagate swirl or very long distances.
Distorted profles and swirl can lead to meters under or over reading. The eects can be large or small and vary rom meter type to meter type. They are known as installation eects as they are dependant on the layout o the pipe system into which the meter is installed.
Flow meters are designed or use in “ideal” conditions but in reality t hey need to cope with valves and bends distorting the ow,, electromagnetic noise, acoustic noise, particles and bubbles, to name but a ew and all o them are potential sources ow o error.
Good Practice Guide
8. Eects on specifc ow meters There are many dierent types o ow meter and installation eects will inuence them in dierent ways. For example some can have large errors caused by swirl whereas some will be relatively unaected by it. Standards and manuacturers quote the minimum length o straight pipe required.
8.1 Dierential pressure meters By dierential pressure meters we are talking about Venturi tubes, orifce plates, critical ow nozzles and cone meters which all work on the same principle in that the owrate is measured using the pressure drop caused by a constriction in the owing uid in a pipeline, a concept frst introduced by Bernoulli in the 18th century. A distorted velocity profle can lead to Venturis, orifce plates, nozzles and cones to over read. With orifce plates, generally speaking, the lower the β the lower the error will be. Swirl on the other hand will cause Venturis, nozzles, cones3 and orifce plates with high β to over read and low β orifce plates to under read.
dd
D D
is of of isthe theratio ratiobetween betweenthe thediameter diameter the restriction (or hole) and the pipe the restriction (or hole) and the pipe diameter. diameter.
Corrosion, erosion erosion and dirt can also lead to errors in dierential pressure pressure meters as they can cause the orifce or throat diameter to increase or decrease. Droplets, such as
Orifice Plate Orifice Plate
the ones ound in wet gas can cause errors but years o research and experimentation experimentation has been carried out on this topic and so correction actors are available. In act, many multiphase and wet gas meters actually incorporate a dierential pressure meter, usually a Venturi tube, into their design although cone meters are becoming more common.
Cone Meter
Putting an orifce plate in backwards, which is a more common occurrence than you may t hink, can cause over reading o up to 20%. Illustrating just how vital correct installation is.
An Introductory Guide to Flow Meter Installation Eects
8.2 Electromagnetic ow meters This type o meter generally requires over 10 diameters o straight pipe upstream or its installation and there can be problems with large bore meters. On the plus side, electromagnetic meters are not as aected by swirl as much as other types o meters are.
Horizontally mounted electrodes
Vertically mounted electrodes
However,, that is not to say that they aren’t aected by swirl at all. Particles or bubbles present However present in a uid can aect an electromagnetic ow meter as they tend to either rise to t he top o the pipe or all all to the bottom. One way to reduce the eect would be to install the electrodes horizontally on the pipe wall.
8.3 Ultrasonic meters The eect a ow disturbance, such as swirl, has on an ultrasonic ow meter is dependent on the path design. As a general rule, the more paths the less o error t here will be as you get a better average o the velocity profle. However, However, the more paths you have the more you will need to pay or your meter. A 4-path meter will be less aected than a meter with 2-paths and an 8-path meter will be less aected that a 4-path meter. meter. When you get up to a 16 path ultrasonic meter you would then need to jump up to say 32-paths to get urther useul inormation and so on. The graph below shows the errors in 3 dierent ultrasonic meters installed downstream o a bend. 4.0 2.0 0.0 -2.0 4 Path 2 Path A 2 Path B
-4.0 single path
double path
double path
four path
nine path
-6.0 -8.0 0
5
10
15
20
25
Distance (D)
8.4 Turbine Turbine ow meters The degree to which a turbine meter is aected by asymmetry and swirl is very much dependent dependent on the design o the turbine blades. I you have swirl turning in the same direction as the turbine blade then it will propel the turbine round aster giving the impression o a aster uid owrate. On the other hand i the swirl is rotating in the opposite direction rom the turbine it can slow the blades down or in an extreme scenario cause the blades to go in the other direction giving a reduced or negative owrate. In some cases swirl generated downstream o the meter can actually have an eect on the turbine. So once again watch out or those double bends.
30
Good Practice Guide
8.5 Vortex meters Vortex meters work by deliberately introducing a disturbance into the ow. This means that other disturbances in the ow can lead to errors in the region o ±3%. Due to this, in some cases, up to 70 diameters o straight pipe could be required which is a signifcant amount o straight pipe. Vortex meters can also lock on to vibrations and pulsation requencies such as those generated when a valve is being opened or closed so that is another actor to take into consideration.
8.6 Variable area meters Variable Variab le Area Meters (VAM (VAMs) s) work on the same principle as dierential pressure meters but instead o having a fxed area and recording the change in pressure, VAMs have a constant pressure and the area changes which is related to the owrate. Like dierential pressure pressure meters they too are aected by ow disturbances and other things such as particles which can erode the oat. They are relatively low accuracy devices even when installed correctly, correctly, ± 2 or ±3% at best.
8.7 Positive displacement meters Positive displacement meters are one o the ew types o meter that are not particularly aected by asymmetry and swirl. However, they can be aected by a change in viscosity. viscosity. I your uid decreases in viscosity and becomes “thinner” then it is possible that some o the uid may slip between the gears rather than propelling them them round leading to your your meter under reading. Also, as positive displacement meters have moving parts in contact with the uid any particles or droplets present can wear away the gears. Even i there are no particles in the uid general wear and tear can occur over a period o time so regular maintenance and inspection is recommended.
8.8 Coriolis meters With a Coriolis meter the uid is directed through a vibrating ow tube, Coriolis orces exerted by the moving uid cause the tube to twist. The amplitude o the twist is related directly to the mass ow rate. Coriolis meters are not very sensitive to swirl and asymmetry but are aected by incorrect installation. When putting a Coriolis meter into your piping system you need to ensure that the vibrating tubes are aligned and not under stress so that they twist and contort otherwise you will get errors.
An Introductory Guide to Flow Meter Installation Eects
8.9 Thermal mass meters There are a number o dierent kinds o thermal mass meters but they all work on the principle in that the rate o heat absorbed by a owing uid is directly proportional to it mass owrate. Since there is such a large variation in the designs no hard and ast rules about this type o meter can be given in this good practice practic e guide. However, good places to fnd inormation are the ISO standard or the manuacturer’s manual specifc or your particular meter.
9. What can be done about installation eects? All is not lost as there t here are ways to reduce installation eects. Ideally the meter would be moved to a more suitable location with the required amount o straight pipe upstream and downstream.. Alternatively downstream Alternatively,, the type o ow meter currently being used could be changed or a dierent design o the same kind o meter could be used i.e. still using an ultrasonic meter but choosing one with more paths. paths. Perhaps even a combination o both would be most suitable.
10. Is there an alternative? There is another solution and that is to use either a ow conditioner or a ow straightener. straightener. Flow conditioners are, generally speaking plates with holes in them and come in various designs. They can remove swirl and asymmetry rom the ow. Flow straighteners on the other hand are usually tube bundles and are capable o removing swirl but not asymmetry. Whilst ow conditioners and straighteners will reduce the amount o straight pipe needed they won’t eliminate the o use straight pipe altogether altogether.. It may seem like ow conditioners and straighteners are the answer to reducing the eect o ow disturbances but it’s not that simple. Yes a straightener will reduce the amount o straight pipe needed by removing swirl but in turn it will increase the pressure losses. A ow conditioner will give even greater pressure losses but will remove both swirl and asymmetry. Pressure losses losses are something you want want to avoid where possible. possible. This is because it takes a certain amount o energy to move your uid rom A to B and pressure losses remove remove some o that energy meaning you need to compensate by adding more energy which ultimately costs you money. These are things that must be considered to see i the trade-o is worth it. Incorrect installation o a conditioner may actually make things worse so take care!
Good Practice Guide
11. Round up So what can you take away rom this introducto introductory ry guide to installation eects? Well, hopeully by reading this, you will now be aware that ow meter error can be increased by valves and bends disturbing the ow as well as by electromagnetic noise, acoustic noise and pulsations, and particles, droplets and bubbles. Dierent meters respond in dierent dierent ways so the type and design o meter you choose is important. Next time you have to buy a meter think outside the anges and consider where it’s going to go and what kind o things it will need to cope with. Take a look at where it’s installed. Is it ater a bend, or worse still a double bend? Can it be moved? Should it be replaced with a dierent design or dierent type o meter completely? Is a combination o both the best solution? Even with all the experimentation and research carried out in connection with installation eects they are still not well understood. This means that standards get updated and manuacturers can change installation recommendations. recommendations. Never design to the minimum, i 20 meters o straight pipe are required ask or 50 and hopeully you will be met in the middle.
Remember, installation eects: Your meter, Your production, Your profts
12. Further reading The ISO standards or meter installations are a good starting point: •
Cori Co riol olis is– –I ISO SO1 107 0790 90
•
Crit Cr itic ical aln noz ozzl zles es– –I ISO SO9 930 300 0
•
Diff Di ffer eren enti tial alp pre ress ssur ure e4–ISO5167
•
Elec El ectr trom omag agne neti tic c– –IS ISO O68 6817 17
•
Positi Pos itive vedis displ place acemen ment– t–IS ISO2 O271 714 4
•
Ther Th erma mal lma mass ss– –I ISO SO1 145 4511 11
•
Turb rbiine e– –IS ISO O2 2715
•
Ultr Ul tras ason onic ic– –I ISO SO6 641 416 6
•
Var aria iabl ble ear area ea– –I ISO SO1 116 1605 05
•
Vor orte tex x– –IS ISO/ O/TR TR1 127 2764 64
Alternatively you could attend one o our training courses to learn more about installation eects and other ow-realted topics.
4
For urther inormation, contact: TUV NEL, East Kilbride, GLASGOW, G75 0QF, UK
Cone meters are not covered by ISO 5167 or any standard or that matter at this time.