bias in Crane 410 fitting losses 1

[gmax137]

Hi all, I'm new to this site. I'm trying to find out if the resistance coefficients for valve and fittings in the Crane 410 paper are biased high or low relative to "reality." I have always figured that they are biased high to ensure that the calculation overestimates the pressure drops, but is this really true? I'm interested in doing a maximum flow calc (to determine if the system losses will prevent pump runout without throttling valves).

Thanks in advance
Gregg

 

[Latexman]

I think some of the graphs in Crane TP410 can give you the bias if you work to extract that information.

Good luck,
Latexman

 

[pleckner]

If you have the time, go through the following thread:
thread378-173164

 

[gmax137]

Thanks pleckner, I read that thread before I posted this one. It is very interesting. I understood the issue there to be about which friction factor to use with the K factors given in Crane (use f or fT). My question is really directed at the K factors themselves - for example, Crane gives K = 8 fT for a gate valve. If we went out and tested a bunch of gate valves, would they average out to 8 or would they all be less than 8? Crane page 2-1 states that the values are based on flow tests, and refers to the values in Appendix A as "representative resistances." I'm still not sure what that means.

Thanks
Gregg

 

[MikeHalloran]

I have the impression that Crane computed their data from actual tests on a sample of representative hardware. That was done quite some time ago. So the data should be representative of hardware as it used to be made.

Some things have changed in the meantime, e.g. most fittings have thinner walls and are probably smaller than is 'traditional', thanks to advances in casting and analysis, and economic pressures to reduce material waste and safety factors.

I.e., pressure drops may be a little larger than Crane predicts.





Mike Halloran
Pembroke Pines, FL, USA

 

[unclesyd]

The numbers in Crane 410 were compiled from actual flow tests. While working for a competitor in the 1950's we had an extensive flow testing setup to test our valves and fittings as well as the competitor's valves and fittings. This was in the days when there was still where cooperation and competitiveness went hand in hand. I remember that we actually ran a series of flow tests for certain Crane valves. At the same time we ran our comparable valve figure to see how close the values were.
From the windmills of my mind I recall that the numbers for comparable valves form all the major players were within all in very small range. This comparison was deemed essential so that a figure/figure component could be interchanged without any extra thought.

 

[IFRs]

Considering the phenominal amount of effort in this and te threads referenced in this post, what are your thoughts on using the "Flow of Fluids" software package that is based on the Crane 410 paper for fully turbulent flow of petroleum oils?

 

[Artisi]

And at the end of the day, is a +- of a couple of percent worth the worry when calculating a friction loss when so many other variables tend to influence the calculation/ selection of pipe lines and equipment to a greater extent anyway.

 

[zdas04]

Artisi,
I had the same question. When you're doing a pipeline project, the uncertainty is typically around 10-20% so does it matter if a gate valve is 8 ft or 18 ft?

I think the real application is in plants where people are sizing pumps/compressors with much smaller safety factors than folks use outside the plant fence.

David

 

[Artisi]

Even within a plant, the margins can be quite variable unless very stricy guide lines are implimented for example;

The process requires a flow of "X" - the process engineer adds a couple of percent margin just in case the calcs are a bit out, the pipeling people add their margin, the "boss" want a bit of marigin so he can get an increased output further down the track, the pump guy adds a bit of margin to the pump - just in case etc etc.
So does the difference between an equivalent head loss of 10 or 12ft loss across a valve / bend / T really matter.

 

[gmax137]

I do agree that the issue of using "fT" rather than "f" for fitting losses, as discussed in the referenced thread, has a small and probably negligible impact on design calculations. But the posters in that other thread also acknowleged that; the concern was getting a proper understanding of the theory. I started this thread while doing some work on already built and operating systems - I have as-built isometrics, valve Cvs and in-situ pump test curves so alot of the design margin issues mentioned above are not pertinent to my situation (though I quite agree with the observations made in the context of design). Just to add another two cents to the discussion, note that the coefficients listed in Crane are given to one or two significant figures. That alone should guide us in the accuracy we ascribe to the results and the level of effort that should be applied to determining the other inputs into the pressure loss calculations.

The original post asked if the Crane coefficients were intentionally biased "high" to ensure that design calculations overpredict the losses, since this is usually the conservative approach. Over the years, several people have told me that this is the case. Based on the responses above, I no longer believe that to be true - instead, the Crane values are probably pretty good estimates. It is therefore up to the user to decide how much conservatism to apply in the calculation.

Thanks to the people who responded.

 

[BigInch]

I've had clients add an extra 15 miles of pipe to "bias" the friction loss on a 150 mile pipeline. A "bit" extreme perhaps, but the point being, if you designed it so close to where Crane "bias" makes a difference, that's way too close. I would say if you hit within +/- 10% of the expected friction loss on a long pipeline, that would be a job very well done.

http://virtualpipeline.spaces.msn.com