70% of installed meters arethe wrong technology or the wrong size. 1

[zdas04]

In 2000, the committee added a detailed section on uncertainty. The outcome of that was that Beta ratio needs to be between 0.32 and 0.70. For a 4-inch tube 0.32 beta is 1.280, so call it 1.250 and accept 0.31 Beta. At 100 psia and 100 [°]F, 50 MCF/d is 0.3 inH20. Since most of these 4-inch tubes are calibrated 0-150 inH20 (standards again) a 0.1% uncertainty is +/-1.5 inH20--at 50 MCF/d you have numbers that have little to do with measured volumes. At 112 MCF/d (1.5 inH20 at 100 psia in a 4-inch tube with a 1.25 plate)you know that you are between zero and 159 MCF/d.

I haven't looked at the graph in a while, but I seem to recall that a 0.625 plate in a 4-inch meter (0.155 beta) would have an uncertainty greater than 4%.

David

 

[hacksaw]

zdas04,

no precise records were kept from the field surveys, paper mill (steam,chemical,water, nat. gas), chemical plants, refineries, water plants, waste water treatment.

generally where the meter and the installaton was an engineeering design, errors in the meter selection and the installation were low, in the 1-2% range. Having said that, in many process designs (non-custody) available piping runs are simply less than ideal, but in those cases acceptible measurement errors were always hashed out with the process designed before hand.

One of the most common flow measurment errors involved transmitter replacements where the transmitter setting were not entered correctly, e.g in the case of d/p meas. using square root extraction in the transmitter when it was already being done remotely.

On the other hand one of the greatest sources or error was in the meter sizing offered by sales organizations as opposed to manufacturers, but in those cases where errors were discovered, it generally involved assumptions of the reference basis (standard T&P conditions for the fluid properties).

Where the meter installations were not engineered, the error rates were much higher since the meters seemd to be purchased on the basis of catalog cuts rather than engineered design. Usually gross mis-application of a meter-such as due to low flow cut-out of vortex meters.

 

[zdas04]

I was wondering about your statement that "70% was high by a factor of 10 to 20". That statement is saying that 93-96.5% of the meters are properly installed and properly sized. The post you just made sounds like you are now thinking that the number of bad meters is more than 3.5-7%.

David

 

[hacksaw]

depends in part on the criteria for "bad" or "improperly sized".

if 70% of the meters were incorrectly installed, operating plants would not function anywhere near design

The installation error rate is more like a 0.1% in engineered plants.

After the plant has operated for more than 20 years you acquire increasing errors, due to process changes, lack of data, meter degradation, transmitter changeouts, etc. These are errors, but not due to incorrect installation,; there again depending on the definition of installation error.

At the time, a 70% installation error across all industries would have been a finacial boom.

As a sales definition, of course all meters are incorrectly installed and need urgent replacement ...

 

[jmw]

For those coming late to the discussion, the link has changed to:

http://www.controleng.com/index.php?id=483&cHash=081010&tx_ttnews

[tt_news]=1782

Looking at a specific application, I wrote an article that explores a possible instance of this where the selection looks like becoming a standard solution instead of one of several options:

http://worldbunkering.com/

click on the winter 2010 edition and visit page 43......

the Coriolis manufacturers probably have a price on my head now

JMW

http://www.ViscoAnalyser.com

 

[zdas04]

Good article jmw. I just wish you wouldn't use the word "accuracy" when you mean "uncertainty" and "repeatability", but everyone does it and it drives me crazy. "Accuracy" just has too much baggage, and everyone thinks they know what it means so it rarely get defined, and everyone's definition is different. Just my pedantic pet peeve.

David

 

[jmw]

Yes, zdas04, guilty.
But with an excuse (albeit an "after the event" rationalisation).

First of all, when it comes to going off topic, and waffling for too long, I'm the prime candidate and in fact the article as published is about revision three, the final version didn't make it in time.

The focus here wasn't really about accuracy except as a parameter by which to compare meters.
If I used terms like uncertainty I would have to explain myself.
When most of the readers (and most readers might find any dissertation on what constitutes "accuracy" misplaced or confusing) get a leaflet from the manufacturers they see "accuracy" statements. They won't actually make this distinction nor think about uncertainty, repeatability or reproducibility nor any other niceties.

What the manufacturers really quote is the performance under factory/laboratory conditions and with some technologies the big differences come when you move away from factory/laboratory conditions.
But again, this would lead way off topic.

There are also a lot of ways to evaluate accuracy many of which may actually not be so technically precise as we'd like.
That is why I referred to different transaction metering examples because I'm quite sure that once away from clearly defined applications, this is something people tend not to consider.

But, far be it from me to suggest I thought consciously too much about this when writing this article. I hope there was some unconscious reviewer at work that did consider these aspects.
I might in other cases make the same choice but without that justification; except that your comment means I will now remember to think about it consciously.

JMW

http://www.ViscoAnalyser.com

 

[jmw]

For some reason the link doesn't seem too good.
The link is all of the following, not just the bit that got underlined before:

http://www.controleng.com/index.php?id=483&cHash=081010&tx_ttnews

[tt_news]=1782

But to be safe the article is:

Flowmeter Challenge: Right Size, Right Design
How many process plants suffer with poorly sized and poorly selected flowmeters? Why does this seemingly simple selection task get so complicated?

Peter Welander, Control Engineering

09/01/2009

But maybe something wrong with Firefox (or me)... sometimes pasting the link works and sometimes not.....

JMW

http://www.ViscoAnalyser.com

 

[zdas04]

Try {link url} My link {/link} (the "{" are really "[") as in =1782]Flowmeter Challenge: Right Size, Right Design

Also, if you think about it even for a second before you write the word "accuracy", you'll be way ahead of the game even if you go ahead an used.

David

 

[jmw]

And the link you posted took me to the wrong article... which is the problem I had.
let's try this:
=1782]Flowmeter Challenge: Right Size, Right Design because somehow, the link doesn't want to include the "=1782". Perhaps because of the [tt_news] in there?

JMW

http://www.ViscoAnalyser.com

 

[zdas04]

It is funny, when I previewed it, it was fine. When I said "submit" it garbled it. The problem is definately the [tt_news]. I think that replacing the closing bracket with the ASCII code {]} might work:
=1782]Try This

Nope, it looks right, but it doesn't take you to the article.

I did {link

http://www.controleng.com/index.php?id=483&cHash=081010&;tx_ttnews

[tt_news{]}=1782
}Try This{/link}




David

 

[ccfowler]

I got to the article through an ordinary search (yahoo, if I recall correctly).

Interesting article, indeed. I've always been fascinated at the great interest in "saving" money by skimping on adequate metering equipment in custody transfer situations. The frugal seller is usually dedicated to a marginally to poorly configured orifice meter that has everything stacked in favor of the customer getting a tremendous gift of "free" product. The lost revenue could easily have paid for a far better metering system over and over and over ...! Pay no mind, tangible funds were saved on the installation.

Valuable advice from a professor many years ago: First, design for graceful failure. Everything we build will eventually fail, so we must strive to avoid injuries or secondary damage when that failure occurs. Only then can practicality and economics be properly considered.

 

[IRstuff]

let's try this:

http://tinyurl.com/Flow-Challenge

TTFN

FAQ731-376

 

[ccfowler]

Thanks, IRstuff! Your link worked on the first try!

Valuable advice from a professor many years ago: First, design for graceful failure. Everything we build will eventually fail, so we must strive to avoid injuries or secondary damage when that failure occurs. Only then can practicality and economics be properly considered.

 

[jmw]

Thanks IR stuff (plus a *).

Sad but true, purchasing cost benefits always win out even if the cost of ownership is significantly higher.
Any tome the bean counters get involved things go wrong.


JMW

http://www.ViscoAnalyser.com

 

[IRstuff]

Well, not always, but close. I usually go for the cheap, but I just spent $99 for an HP laser printer, on the hope that it will live as long as my dearly departed HP LaserJet 5P, which lasted a solid 15 years, and only quit a couple of days ago. It's just spitting out blank pages, even though the demo page prints out correctly. It will be missed.

While $99 is relatively cheap, the Samsung was listed for only $49...

There are a few instances in consumer products where products with stellar reputations can and will survive a large number of onslaughts of cheaper products. Of course, once their reputations get sullied, all bets are off.

TTFN

FAQ731-376