Water Meter Accuracy Measurment

Published on by in Technology

How is the accuracy of water meters tested?

Was it proven or is it convenient for utilities to adopt the volumetric meters and retain the inferential meters which are simple and cheap instead of the expensive meters which show the real consumption?

The idea of equability is that volumetric meters (oscillating piston and nutating disc) and the inferential meters (single jet and multi-jet) will show a registered consumption that is not the true consumption but that by the equable measurement.

The registered consumption can be used as a proxy for the true consumption without any consumers being significantly advantaged or disadvantaged by the reading.

For example, if the meter registers 100 m3 then the unregistered consumption may be 5 m3 and it will be, within acceptable limits, the same 5 m3 unregistered consumption for all consumers.

Taxonomy

12 Answers

  1. Technically, you need to consider the following factors in determining water meters which is also related to its accuracy and capability of measuring "real consumption":

    The first is to determine your average/operating flowrate on your installation (commonly known as nominal flowrate or Q3). You need also logged data to know the flow profile of the installation, which flowrate consumes the most volume and spends most of the time.

    The second is the Ratio (Q3/Q1), where Q3 is the average flow rate (operating flow rate) on the installation with +/- 2% of error margin and Q1 is the minimum flow to be measured by the water meters to be between +/- 5% of the error margin. Note that starting flow rate (Q start) is another point of measurement which is below Q1 with no error margin (could be more than +/- 5%) and not included in the metrological error of measurement. The bigger the Ratio (R), the better the measurement range and its capability to measure lower flowrate. For example, common mechanical meters have a Ratio (R) between R50 - R160, but current solid state meters could be up to R300 - R400 which is better to measure low flow rates.

    The third one is the "Typical Error Curve". The typical error curve is unique for each meter which depends on the design, operating principle, materials, and many other technical factors. Here, you need to deeply analyze which flowrates operate the most and its typical error curve on that flowrate. For example, if 99% of your consumption is resulted by flowrate on Q3, then you need to find a meter with near-zero percent error on Q3 (or adjust the measurement to near-zero percent of error because some meters allow you to adjust the measurement).

    Last but not least, current solid state meters (ultrasonic, electromagnetic) have enhanced their capability of measuring constantly near zero percent (+ 0.2% for example) which means the meter error is constantly between + 0.2% between all flowrates (no undulation of typical error curve like mechanical type meters).

  2. Andy Godley :

    My point here is that:

    PD meters (e.g. reciprocating piston) Complex, large, expensive but where the registered consumption = true consumption. This kept them in the market from 1824 till sometime in the 20th Century (1934? the IMO bi-rotor patented and "the most accurate water meter ever")

    Inferential meters (direct plumbing): simple, compact, cheap  and, in direct plumbing when first introduced, the registered consumption = true consumption but in indirect plumbing a very significant difference between registered consumption and true consumption.

    The rotary piston meter (semi-positive, patented in the 1860s) is simple, cheap, compact and, in direct plumbing the registered consumption is the true consumption but in indirect plumbing there is still a significant difference between registered consumption and true consumption.

    In the 19th cent. inferential meters were referred to as "rotary" meters and hence direct plumbing as the "rotary" market. Tylors,introduced the "rotary piston" meter as the "British Patent Rotary water Meter". It competed with the inferential meters due to its lower flow rates (and other weaknesses of inferential meters) but not then the PD meters in the indirect plumbing market. Tylors continued to evolve the reciprocating piston meter (new design patented in 1888) for indirect plumbing.

    Unchanged since the 1870s (Nash patented improvements) while PD meters continued to evolve

    at some point in the 20th cent. Rotary piston finally displaced all the PD meters.

    All i see as different is the "Equability" concept which I see appears in the 1942 water act (and in many standards since) This enables the rotary piston meter to take over from PD meters despite the registered value not being the true consumption.

    My question is about the validity of the Equability Concept, how it is defined and tested.

    Or is it just a convenient fiction that allows the simple cheap compact meters to take over from the expensive complex bulky PD meters which otherwise should ahve been dispalced in the 1860s or 1870s?

    PS Static//Solid state meters also depend on the Equability concept. More so, perhaps than is realised.

    1 Comment

    1. Regarding equitability, there are discussions in the various water meter standards committees about how to implement the requirement in the Measuring Instruments Directive (MID) that "The meter shall not exploit the MPE or systematically favour any party" in a meaningful test. As meter error curves are a function of flow-rate, should this be based on a typical consumption pattern (how should that be defined as there is so much variability?) or is it over time (what time period etc). It's not an easy question to answer.

      Quite frankly, as far as meter technology is concerned, I don't see us turning the clock back 150 years and starting again! However, if you have a meter that provides improved accuracy, particularly at low flows and is economically viable and meets various other criteria outlined on the previous post,  then we would be delighted to include that in our next test programme. We have used pd meters as reference meters on a number of our test rigs - they are mechanical and do wear - in who's favour? 

      I think we need to be looking at some of the questions that are being raised on solid state meters and ensuring that these are captured within the tests in the various standards for metrological approval. That the meter should not systematically favour the customer or water provider is an important aspect of this.

      1 Comment reply

      1. From 1824 on the search was for a PD meter that is simple, compact and cheap. That hasn't been found. Yet. The volumetric meters have been a near miss being only semi-positive.

        Calibration shift and the effects of wear are a function of the slip flow. The lower the slip flow the less the effects of wear.

        The reciprocating piston meter today, with its flexible seal between piston and cylinder, and in its core applications (petrol and diesel), will operate for 12milliom litres with no "significant" change in accuracy... and in that application "significant" is probably far less than for a water meter.

        The volumetric meters probably have very little clearance flow but suffer because of the open flow paths (through the piston wall or disc) created over 20-30% of the cycle, are responsible for significant unregistered flow, the non-linearity etc. and thus vulnerable to wear or anything else that affects "slip flow".

        I do think solid state meters are perhaps misleading as to the significance of claimed advantages. For example, the lower flowrate. If it was a volumetric meter with such a low flow it would be fine, a clear advantage but when one takes into account the blockage  ratios of the different technologies volumetric meters affect the native flow profile while the solid state does not. The PD meter, for example, below the starting flow rate stops registering but it also stops the flow. In the volumetric meters below the starting flow the flow rate through the open flow paths is significantly reduced. In solid state meters the flowrate below the starting flow i only marginally less than the starting flow.

         

  3. I'm not quite sure of your point here. The choice of a meter for any application is always a compromise between many different factors - including performance, cost, reliability, expected service life etc etc. The rotary piston meter we tend to use in the UK has evolved as the (current) meter of choice because it does a pretty good job of balancing these compromises. It is relatively cheap, will last typically 15 years (admittedly with some degradation of performance but that happens with any mechanical meter), is reasonably robust, it has good low flow sensitivity (which is important where much water supplied is stored in tanks with ballcocks to control inflows) but also has a wide measurement range  - wider than most other meter types. This minimises the number of different sized meters required to cover properties with different consumptions, making installation simple. Importantly the impact of mechanical wear tends to be in the customers' favour, i.e. the meters tend to under-record over time. This means that there is no need for regulation around replacement tor expensive testing programmes driven by consumer protection that there is in countries that use technologies that are likely to over-record. We also like putting meters in boundary boxes underground and outside - few other meter types have been developed as manifold fitting.

    Whether these considerations hold up as solid state meter types begin to penetrate the market remains to be seen...

  4. I work for the government water and wastewater utility in our province here in Canada and we use 2 meters only.  Sensus meters on pipelines 5/8" to 1" (outside diameter) and magnetic meters (specifically Rosemount mag meters) for flows on pipelines exceeding 1" up to 36".  We have spent much time and resources researching the best meters for our customers and after years of testing we have decided that the mag meters are the only way to go for high flows and the Sensus meters for the small flows.

    The accuracy of the mag meters that we use have a volumetric accuracy of ±0.25% with a turndown range of 13:1 to 40:1 with a ±0.5% accuracy.  These are very high quality meters.

    The Sensus meters for the small flows (5/8" to 1") the accuracy is ±1.5% on normal flows and on low range flows the accuracy is ±3%.  Normal flows = 0.04  m3/hr to 12.5 m3/hr and low flows = 0.025 m3/hr to

    1 Comment

    1. Dear Kurt, have you checked on a patented magnetic flow meter from KROHNE named Waterflux? You'd like to evaluate that as the turndown range is 400:1 is it is optimum even for low flows (typically during night). Kind regards, Claudio Costa

  5. Igor Šušić "There is no water meter that is 100% accurate".

    Not now there isn't but the 1824 reciprocating piston meter is, or became, the most perfect example of the most perfect technology - positive displacement. reciprocating piston has a flexible seal between the piston and cylinder so there is virtually no slip flow, just what could pass through the four way valve.

    As positive displacement meters the only way for flow to pass was to move the piston. When there was not enough pressure drop to move the piston the piston stopped and flow was stopped by the piston.

    Positive displacement is the principle  behind ball and piston meter provers, the Bell prover for gases and the flying meniscus and flying bubble methods for low flows. They are used to calibrate all types of meter.

    Between 1824 and 1871 over 300 new meter patents were awarded in the quest for a simple cheap and compact positive displacement meter. Many were improvements to existing designs.

    The PD meter registered consumption was the truer consumption. The reciprocating piston meter today is virtually unchallenged for its performance.

    Rotary piston and nutating disc did not advance after the 1880s. New POD meter designs came such as Tylors reciprocating piston of 1888 and the IMO birotor of 1934 - the most accurate water meter ever.

    Then PD meters lost the market to the unchanged rotary piston and nutating disc. The only reason I can see is the "Equability" concept. This allowed the volumetric meters to be judged as giving a reliable proxy measurement. This appears in the UK 1942 water act. It allows the simple and cheap volumetric meters and the inferential meters to take over the market. Solid state are also only proxy meters.

    This is my question: How is equability tested and verified? (Someone changed to title of my original post). 

    Of course, with a true PD meter as simple and cheap as the volumetric meters, the whole market could change again and just as dramatically. That the PD meters are complex and expensive is simply history. Indeed, the Lewis Nash Crown meter could ahve been that. mechanically simple that it has complex geometry was the problem but the Kent NSM innovation of the 1960s could have solved that problem and delivered a simple cheap true PD meter.

  7. I'm not sure I made my problem clearly enough.

    When using a meter to bill for consumed water it is important that the meter registered value is the true consumption. The consumer will be billed at a unit cost based not simply on the water consumed but also for all the associated costs.

    The inferential meters, first used for direct plumbing, satisfied this requirement because, with a 1/4 turn tap on the incoming mains, the native flow profile was of bursts of flow at maximum flow. There is no significant flow below the meter's starting flow rate.

    The positive displacement meters were necessary for indirect plumbing where a gravity tank is maintained full by a float valve. This means that while a lot of flow takes place at high flow rates, a significant proportion of flow takes place at low flows diminishing to zero. The inferential meters cannot handle this but the reciprocating piston meter (Thomas Kennedy, 1824) does. Not because it measures to zero flow but because, when the meter stops, flow stops and there is no significant unregistered consumption.

    The volumetric meters (Rotary piston introduced in the 1860s) and nutating disc (Introduced in the 1880s) are only semi positive. This means that when they reach their starting flow and registration stops, the open flow paths allows some flow to continue unregistered.

     

    This is not therefore about the accuracy of measurement within the operating flow range of the meters but about what happens below the starting flows. A volumetric meter which registers 100m3 (+/-allowed accuracy) may actually represent a true consumption of 105M3. Equability simply means that it is the same relationship for all meters and thus the registered value can be used a a proxy for for the actual consumption.

    Before this concept as introduced the volumetric meters were intended to extend the measurement capability in direct plumbing in competition with inferential meters (and perhaps why inferential meters are not used in direct plumbing systems in the principles countries of manufacture (UK and USA).

    The volumetric meters have not significantly improved since the 1880s (the last of the Lewis Nash patented modifications) nor made any commercial advances, not till Kent introduced net shape manufacturing in the 1960s.

    While the simplicty and low cost of the volumetric meters was undoubtedly attractive, irresistibly so, PD meters seem to have carried on into the 20th century.

    My point is that the unchanged volumetric meters were unable to displace t

     

  8. I'm not sure I made my problem clearly enough.

    When using a meter to bill for consumed water it is important that the meter registered value is the true consumption. The consumer will be billed at a unit cost based not simply on the water consumed but also for all the associated costs.

    The inferential meters, first used for direct plumbing, satisfied this requirement because, with a 1/4 turn tap on the incoming mains, the native flow profile was of bursts of flow at maximum flow. There is no significant flow below the meter's starting flow rate.

    The positive displacement meters were necessary for indirect plumbing where a gravity tank is maintained full by a float valve. This means that while a lot of flow takes place at high flow rates, a significant proportion of flow takes place at low flows diminishing to zero. The inferential meters cannot handle this but the reciprocating piston meter (Thomas Kennedy, 1824) does. Not because it measures to zero flow but because, when the meter stops, flow stops and there is no significant unregistered consumption.

    The volumetric meters (Rotary piston introduced in the 1860s) and nutating disc (Introduced in the 1880s) are only semi positive. This means that when they reach their starting flow and registration stops, the open flow paths allows some flow to continue unregistered.

     

    This is not therefore about the accuracy of measurement within the operating flow range of the meters but about what happens below the starting flows. A volumetric meter which registers 100m3 (+/-allowed accuracy) may actually represent a true consumption of 105M3. Equability simply means that it is the same relationship for all meters and thus the registered value can be used a a proxy for for the actual consumption.

    Before this concept as introduced the volumetric meters were intended to extend the measurement capability in direct plumbing in competition with inferential meters (and perhaps why inferential meters are not used in direct plumbing systems in the principles countries of manufacture (UK and USA).

    The volumetric meters have not significantly improved since the 1880s (the last of the Lewis Nash patented modifications) nor made any commercial advances, not till Kent introduced net shape manufacturing in the 1960s.

    While the simplicty and low cost of the volumetric meters was undoubtedly attractive, irresistibly so, PD meters seem to have carried on into the 20th century.

    My point is that the unchanged volumetric meters were unable to displace t

     

  9. The new trend to calculate the accuracy of the meters It is the meter error registered in the test bench against the average consumption histogram, obtained with the data logger installed in the sockets. The consumption histogram and the meter error are evaluated for the different expenditures especially the movement and start-up; The idea is to have as many expenses as possible from the meter's performance curve. With this relationship the actual meter volume is obtained for each meter. 

  10. Accuracy is what the manufacturer has to demonstrate before getting its method and instrument approved for a requested class of precision (accuracy). Accuracy classes are defined and used in IEC and ANSI standards.

    The engineer opts for a class, and afterwards for the instrumentation capable inside the class.

    The accuracy of the measurement on site will be subject of respecting basic rules of engineering after having a method implemented. Errors might come from external factors like random variation of light, humidity, pressure, just name it.

    Accepted tolerance has to stay inside the accuracy class ( if we wanted to measure 1000m³/h within a tolerance of +- 0.1% then we need to be able to have the accuracy of class 0.1 by using an instrument which was certified (and calibrated accordingly), where 1m³/h will be the margin of error (plus or minus), resulting in a range of uncertainty  of 2m³/h... 

    Usually a measuring instrument has to be 3-10  times more accurate then the lowest tolerance margin (10 times more accurate for metrology calibrations, 3 times at least in op field). Many times these rules are skipped for the sake of price savings. 

    Coming now to the two methods... volumetric and inferential.
    See this link and its subclasses.  The complexity of the PD makes them expensive, accurate and kind of reliable. The others may be price convenient, but if they must fit the same accuracy class, they will get complex and expensive, too. 

    The trend is to minimize the gap in price using latest achievements. Unfortunately, many public companies using public funds cannot afford the risk of experimenting, and they are conservative. (not all of them :) )
     

     



     

  11. The accuracy of flow meters is achieved though weight table measurement. Water is weighed and this gives a true value for the volume. National Physics Laboratory  is one body that offers such a facility. I'm sure there are Standards (BS, ISO etc) that meters have to comply with. Water companies change out meters on a planned basis to account for when inaccuracies might begin to occur. Accuracy is flow dependent for different types of meter. Meter under-registration is a well documented phenomenon. When undertaking studies in different parts of the World into things like leakage, engineers would often take a well calibrated meter to assess the validity of local meters. Hope this answers? 

  12. Dear Mr. Watson. You can take a calibrated bucket like 200L oil bucket or something similar and fill it while registereing metered volume. There is no water meter or water metering procedure that is 100% accurate. Filling a pre-determinated volume is only accurate way. Good luck...

    1 Comment

    1. I think we are talking about two different things here.

      Accuracy is often considered as the accuracy of the measurement within the meters operating range. In most transaction metering applications this is important, in some such as fuelling your car, critically important. However, in utility metering the unique circumstances are such that it doesn't matter much if the meter is (+/-)0.1% accurate of reading or (+/-)2.0% or reading accurate.

      However, that accuracy may be a perceived quality factor. On the other hand what is important is the accuracy with which the meter registration reports the true consumption and because no meter in use today can do this, the meter reading is a proxy for the true consumption but is required to be "equable". That is, no one consumer should be significantly advantaged or disadvantaged by the reading.

      My problem explaining this or getting a response is that terminology changes. "Equability" was the term used by my mentor at Tylors when explaining the necessity for this when the end came for PD meters (due, he said, to the 1945 Rural water supplies act in the UK).

      This is today referred to as bias e.g. in this:

       

      "2006 No. 1268WEIGHTS AND MEASURES

      The Measuring Instruments (Cold-water Meters) Regulations2006

      "(4)The  errors  of  a  cold-water  meter  at  flows  outside  the  controlled  range  shall  not  be  unduly  biased. ""

      In other words, the unregistered flow should not be significantly different fro different users. It is of little value to have a meter capable of (+/-)0.25% accuracy of reading when the error due to unregistered flow can be 5% and vary from 0% to 10% across different users.