March/April 2012
Your Letters

Starting with the correct consultant

In the November/December "Talk to Me", Bill Lydon commented that, "Hiring expensive consultants ... (to) recommend ways to improve your plant should be the last place to start" and suggested the reader start with people in the plant. Having worked both in the plant and as a consultant, I respectfully disagree.

While working in a plant, a consultant suggested plant modifications that significantly increased capacity while dramatically reducing the number, duration, and cost of outages. We (at the plant) may have had tribal knowledge, but we did not have the consultant's broad exposure to other plants, industries, and technologies. Simply put, the value of the consultant's work greatly exceeded the amount of his invoice. Was the consultant expensive? No.

As a consultant, my specialized knowledge of flow measurement often enables me to unravel problems that people at the plant cannot resolve. In one case, the operation of a water system in a major U.S. city "calmed down" (with reduced maintenance) after implementing a control strategy developed to improve flow measurement. The value of my work exceeded the amount of my invoice for this work. Was the consultant expensive? No.

I suggest that the company that hires the correct consultant at the correct time to do the correct work in the correct quantity will survive and prosper. Conversely, companies that fail to use consultants effectively will not reach their full potential. The correct consultant will tell management if the people in the plant are the place to start and will help plant people gain credibility. If people in the plant are not the place to start, the correct consultant will make suggestions for improvement. With most clients, I find myself doing both-helping plant people gain credibility and making suggestions.

IMHO [in my humble opinion], consultants are not the "last place to start." To the contrary, the correct consultant should be the first place to start.

David W Spitzer, Spitzer and Boyes, LLC

Orifice plate measurement

The January/February "Sizing orifice plates" article purports to show that the effects of potential orifice plate measurement errors can be reduced by using a higher full scale differential pressure. In other words, a (say) 1"WC measurement error exhibits a smaller flow measurement error in a 200" WC full scale than in a 50" WC full scale. The article suggests new rules of thumb that "assume an imperfect installation" and "consider using 200 inches as a default, rather than 50 or 100 inches." Further, it is suggested that the graph be used to "gauge if expected accuracy and turndown are satisfactory."

I leave it to the reader to decide if the information presented in this article is complete, correct, and not misleading.

Some comments:

  1. Using a higher full-scale differential pressure (at best) reduces what is likely to remain a significant flow error. From a practical perspective, the error should be mitigated by correcting the installation or via calibration (when appropriate).
  2. The photograph (Jan/Feb InTech, page 33) depicts an "orifice plate installation for gas flow. The low point … can be expected to trap liquid and cause a small measurement error." However, this is not necessarily correct because the low point may or may not be expected to trap liquid. Many common gases (including nitrogen and instrument air) are dry, do not trap liquid, and can sometimes promote evaporation.
  3. The graph represents an installation where the measured differential pressure is nominally 1"WC higher than the actual value. However, the measurement error could also be nominally 1"WC lower. The flow errors associated with positive and negative measurement errors can be significantly different-especially at low flow rates. For example, the flow errors associated with positive and negative measurement errors of 1"WC at 10% flow with a 200"WC full scale are 2.25 and 2.93 percent of full scale flow, respectively. Therefore, the graph is incomplete and should depict the worst case error, or both positive and negative errors.
  4.   Error and performance statements are typically expressed as a percentage of the actual flow rate or as a percentage of full scale flow rate. Information about these expressions can be found in Chapter 3 of my book Industrial Flow Measurement, Third Edition (, Chapter 4 of Flow Measurement, Second Edition (, and at
    • a. Errors are expressed as percentages throughout the text. Citing a percentage without stating whether it refers to flow rate or full scale (or something else) is incomplete, meaningless, and can easily mislead the reader. 
    • b. The caption under the graph states that the error "is expressed as percentage of full scale." Notwithstanding the comment immediately above, the reader has little choice but to assume that all of the percentages in the article are also percentages of full scale.
  5. Using the assumption immediately above, the article implies that a 1"WC measurement error with a 200"WC full scale will perform within a (satisfactory) criteria of 2 percent of full scale (common of orifice plates) over a 10:1 turndown. 
    • a. At 10 percent of full scale flow, the positive and negative flow errors are 2.25 and 2.93 percent of full scale respectively-neither of which meets the 2 percent of full scale (satisfactory) criteria. 
    • b. The article incorrectly characterizes orifice plate measurement systems as 2 percent of full scale systems when they are actually 2 percent of flow rate systems. 
    • c. The 2.25 and 2.93 full scale flow errors correspond to 22.5 and 29.3 percent of flow rate respectively-both of which are over 10 times higher than the 2 percent of flow rate criteria. 
    • d. The actual turndown is approximately 2:1 in order to stay within a 2 percent of flow rate error. This is much lower than the 10:1 turndown stated in the article. With such a limited turndown, the flow measurements will likely be virtually useless over much of the flow range. 
  6. Graphing the error using the full scale error (2.93 percent) instead of the flow rate error (29.3 percent) can easily mislead the reader. A 2.93 percent of full scale error is not good, but it does not sound all that bad. However, its 29.3 percent flow rate error clearly indicates poor performance. In most cases, flowmeter performance should be evaluated using percentages of the actual flow rate. 

Do you plan to use the new rules of thumb? I do not. 

David W Spitzer, Spitzer and Boyes, LLC


David Spitzer adds a lot to this topic. However, I do not necessarily agree with all his points:

  1. The article details significant improvement in error and turndown, especially at low flow rates. Achieving an ideal installation is not always practical.
  2. In practice, almost any gas may be dry or wet. That is one reason installation guidelines are the same regardless.
  3. The error from reading 5% when it should be 6% is the same as the error from reading 6% when it should be 5%. I don't think this is an important distinction to pursue further.
  4. 4-6. All of the numbers in the article are accurate, as is the graph. As Spitzer says, using percent of full scale is a common industry method, and as he also points out, the graph is clearly labeled. In writing the article, it wouldn't have necessarily helped to provide a second set of numbers based on percent of actual flow, as he has done, except to further emphasize the point, because the error grows even faster at low rates if percent of actual flow is used, as his numbers highlight.

I think it would be helpful to industry as a whole if Spitzer addressed this topic in the next edition of his book, so that others may gain from our experience, which was the point of the article in the first place. The article includes some excellent information that I have used successfully in practice for many years already, and I have received a number of positive responses from people who do plan to consider the new rules.

Allan G. Kern, P.E.

Editor's note:

As common as flow measurements are using orifice plates, there are various thoughts regarding design, application, rules of thumb, and field practice. InTech staff reviewed this with other industry experts, and there are varying thoughts on this topic. This would make a good topic for a session at the ISA Automation Week 2012 for anyone that is willing to submit a paper. InTech invites other thoughtful insights on the subject.

Having a helping hand

I read your interesting "Talk to Me" article in the November/December InTech. I would like to share some remarks. From my short experience in both consulting and as an industry expert, I have noted that sometimes it can be helpful to have someone take your hand to cross the road, if I may say so. In that case, a consultant, who helps formulate issues, results in the correct way, and also raises funds, is indeed very helpful.

One communication professor used to say one cannot be at window and see himself walking outside. This requires careful observations, questioning, discussions, and steering with the insiders.

Another reason for having consultants' insight and inside the company is there may be conflicts between professionals inside the factory, not to mention other interests, such as being friendly with certain vendors.

I do agree that having vendors diagnose the problems may be biased.

However, my point of view may be because of origins and the people I am meeting. I am looking forward to reading your comments.

Pousga Kabore