Maximum velocity is a frequent topic on these forums.  There must be some widely varying assumptions to get the range of numbers that I've seen here.  

For "dry gas" flow I use an actual velocity (as opposed to "velocity" calculated as SCF/FlowArea) of 100 ft/second or 15 psi/mile pressure drop as my upper design condition.  Design capacity will be the lower of the SCF/day you get from 100 ft/sec or the SCF/day you get from 15 psi/mile pressure drop. That choice stays way below errosional velocities, won't slam a slug of water into plant-inlet pipeworks, and gives me a makeup compression Hp requirement I can live with.

All of the vertical flow correlations show a critical velocity (for keeping water mobile) around 36 ft/sec.  I figure that if it works in vertical flow, it should be quite conservative in horizontal flow.  I've been using 36 ft/s as a minimum design condition for about 7 years now and see considerably less liquid accumulation in the piping I designed than I see in the piping I inherited.

Hope this helps

David

There is a good thread on maximum velocity in this room with the title "what is the maximum velocity a pipe can take", posted on 8/20.   A good rule of thumb that I have been using is the manufacturers warranty on valves.  When I used certain valves on water service, it turned out that they were rated for no more than 11 fps at which point you were on your own with no support from the factory.  Thats not to say you cant move water quicker, but why would you want to, the energy cost of doing this would be far greater that the cost of piping or equipment in the long run I think...

The minimum velocity I design around is that which will provide reynolds numbers in the sufficiently enough in the turbulent range that I would not find Re's in the transition or worst yet, laminar flow regeims when the system is working.

I hope this helped..

Bob  

I will first answer regarding the minimum flow velocity in a pipe.The minimum velocity is generally dictated by economic considerations; if you take a lesser velocity you end up with a bigger pipe size.The other considerations would be fouling characteristics of the fluid.

Regarding maximum velocity for erosion would depend on type of liquid and duty- short duration operation or continuous operation.Obviously, if it is a short duration/cycle operation, a higher velocity could be used. For pump suction lines and liquid near its vaporization point, maximum velocities are limited.For noise and especially for control valve applications, talk to the control valve suppliers fot the maximum recommended velocity.

Hope this helps.

Arun

One thing to keep in mind is the effect of velocity of check valves, if applicable.  To maintain a check valve fully open, without wear, a minimum velocity is required based on the specific valve type.  Swing checks are the most susceptible to damage if the disc is constantly oscillating in the flow stream and require 12-18 feet per second to fully open.  Nozzle Checks fully open as low as 3 feet per second.

pumpvlvguy point is very important. Crane's TP #410 gives a formula for each tipe of checkvalve... the symbol U/L in crane's paper means : Underwriter's Laboratory (not defined in the Nomenclature page).
Saludos.
a.

The 1997 ASHRAE Fundamentals Handbook page 33.3 table gives following information:

Max water velocity to minimise erosion

Normal operation             Velocity
hours per year                m/s
1500                          4.6
2000                          4.4
3000                          4.0
4000                          3.7
6000                          3.0

Same page also says "velocities on the order of 3 to 5m/s lie within the range of allowable noise levels for residential and commercial builings"

You can check the following threads for more valuable comments. I have made the same search recently..
Try advanced search/check "exact phrase" and write the below subjects, the threads will come up..Good luck

FORUM: Piping & fluid mechanics engineering
SUBJECT: Velocity limitation in pipe

FORUM: Chemical plant design & operations
SUBJECT: Maximum velocity of fluid in pipe

A possible reason for a minimum flow not mentioned so far is that you may want to keep horizontal pipes full to to minimize pipewall corrosion or to preclude a stratified flow condition that would allow entrained air or gases to collect in the water void space and perhaps aggravate waterhammer effects on quick closures of valves. I found an old equation that might be used to estimate the flowrate to keep a horizontal pipe full. It's in a Design News reprint #RS528 from the June 12, 1963 issue of that magazine and is titled "Fluid Flow from Partially Filled Pipes" by D.P.Costa. It is:
    q=7.3d^2.56*K^1.84 where K is the "fullness" fraction from 0.2 to 0.6, d is inner pipe diameter, in inches limited to 2 to 6 inch pipe and q is GPM. The equation came from a 1928 Purdue Univ. bulletin. I tried the equation on a much bigger pipe of 11.7 inch ID and ran the fraction up to 1.0. It didn't blow up and gave rather believable results for full flow. Has anybody got something better to calculate minimum flows for full horizontal pipes?

To Pumpvlvguy and Abeltio,
Regarding swing check valve pivot hardware degradation from too low flowrates to backseat, there's an obscure report that deals with the influence of upstream pipebends on valve disk dynamics. It's Kalsi etal (1988), "Prediction of Check Valve Performance and Degradation in Nuclear Power Plant Systems", Kalsi Eng. Rpt. #1559 (NUREG/CR-5159). Among the findings:
  "At proximities of 3 to 5 pipe diameters, upstraem elbows require an increase in flow velocity over baseline of 10 or 15% to fully open the valve disk. Proximities of 0 to 1 diameter will require up to 50% higher flow velocity over baseline to fully open the disk, except for clearway designs which will require velocities more than 100% higher than baseline [clearway disks lift nearly totally out of the flowstream and may never achieve stability under some flow conditions]."
   "The amplitude of disk motion which occurs before fully seating the disk increases as the flow disturbance is brought closer to the valve. The maximum disk fluctuations in the case of a severe turbulence source can reach as high as 16 degrees and, for elbows, up to 9 degrees. The reducers have negligible effect on disk fluctuations."
   "The onset of tapping begins at lower flow velocities when the elbow is oriented up than when oriented down. This results in wider tapping zones for elbow-up installations compared to elbow-down. This is particularly evident at 0d and 1d and less so at 3d and 5d (d is upstream pipe diameter distance)."
   Kalsi etal appear to be talking about single upstream bends. Multiple, closely spaced bends upstream of swing check valves could have considerably worse consequences.

Hi Vanstoja,
 Your formula is very useful,But "d" has dropped !!
 And also give a small example with your formula.
                                   Thanks
 P.S. for full k=1 ?

I don't understand what you mean by "d has dropped". Full is K=1.0. I ran 2, 6 and 11.688 inch diameters in an EXCEL spreadsheet and got the following results:
  ID(in.) K     Q(GPM)
   2      0.2   2.23
   2      0.4   7.98
   2      0.6   16.82
   6      0.2   37.09
   6      0.4   132.79
   6      0.6   280.02
  11.688  0.2   204.47
  11.688  0.4   723.03
  11.688  0.6   1543.60
  11.688  0.8   2620.74
  11.688  1.0   3951.28
The increased flows for 11.688 pipe at K=0.8 and 1.0 are part of a progression of step increases of approximately 500, 800, 1100 and 1300 GPM for 0.2K steps. The 0.8 and 1.0 K flowrates are "reasonable" though I can't say for sure that they are correct. The test results were apparently only run up to 0.6 and the author was reluctant to extraapolate any further.

There is an ASTM Standard on maximum velocity of hydrocarbons in pipelines, sorry i don't recall the number but a search on their web site will find it. The maximum velocity here is predicted by static electricity build up.