Following my analysis of the data on Skeeter's experiment (posts 254 & 335), my verdict in short is that the pump and experimental rig cannot be faulted for any inconsistent behaviour, based on the fairly exhaustive information provided. So I must now take back my earlier accusations!
It should of course be understood that such situations do not permit any precise analysis. Some 20 to 30% variation between theory and practice can be considered acceptable, because a large number of sometimes questionable assumptions may have to be made, apart from measurement errors and doubts regarding the validity of equipment specifications.
Nevertheless let me try to present my findings in a manner which I hope will be of interest to at least a few CR4 readers who haven't yet given up on this thread.
Assumptions regarding the experimental set-up
The pump is a self-priming centrifugal type with radial flow impeller. Connection size is not indicated, but I have assumed that 'garden hose' of 5/8" nominal size having 16 mm ID has been used for inlet and discharge.
About 1 m of hose has been considered at the inlet, without any foot valve or strainer. In the discharge line, a Y-hose coupling connects to a pressure gauge. What the pressure reading represents is a bit uncertain, as a part of the velocity head may also get measured. A 'side' tapping or tee branch would have been preferable, but this is a minor issue.
Since the main container, pump and bucket are at about the same level, it is assumed that the pump does not have to overcome any static head (level increase from suction pool to discharge point).
One bit of information I have ignored is the discharge hose going over the edge of the large container. I presume this is a physical layout constraint -- maybe the pump was near a wall and the container had to be placed in front of it, which is anyway convenient for the 'open discharge' test. (Any possibility of the pump being placed inside has not been considered.)
Assumption relating to the calculations
Data about the pump itself is rather meagre. Rated flow is mentioned as 80 lpm, which probably refers to a state of free discharge from the outlet. Maximum head is given as 7 m water column (abbreviated as mWC hereafter), but the corresponding flow rate is not indicated.
Motor data is missing, but my guess is that it would be a two-pole cage motor which at 60 Hz supply (in the US) would run at about 3400 rpm. Power rating may be 1/4 or 1/3 HP as a rough guess, considering that a self-priming pump typically incorporates a vortex chamber and will have a low hydraulic efficiency because of the complex flow pattern within the pump casing.
Calculation of flow losses in pipes requires a whole set of assumptions. Kinematic viscosity of water is 0.8 cSt (or mm2/s) at 30°C. The hose has an internal roughness of 0.3 mm, for which the Colebrook chart yields a friction coefficient of about 0.05 for turbulent flow in rough pipes at Reynolds numbers over 45000 or so. Some of these values could be questioned. By the way, the French apparently have a name for the SI unit of dynamic viscosity -- what else but Poiseuille!
Results
I am skipping all the tedious working and presenting just a summary (do I hear sighs of relief?). If perchance somebody checks it all out and arrives convincingly at substantially different conclusions, I'll agree to any reasonable form of penance. (My difficulty lies in finding someone who is willing to cross-check my work -- it is a tricky business, as pointed out in the Projectiles thread, posts 60 to 62.)
In the absence of adequate pump data it is not possible to calculate what the flow rates 'should' be. What is presented below is a selection of parameters which I feel are relevant, corresponding to the measured flow rates of 30 and 40 lpm, plus 80 lpm assumed for free discharge. The working is in the metric system, with a few fps equivalents.
For brevity in presentation, the three numerical values for each parameter are just separated by a slash.
The correlation between the last two lines can be taken as reasonable, considering all the uncertainties and assumptions involved.
Concluding Remarks
Reducing the flow losses in the discharge line from a radial impeller pump (or blower) for increasing the flow rate should not be resorted to blindly. If the unit is already delivering its maximum rated flow (usually difficult to measure in practice) any further increase could lead to motor overload (tripping or burn-out). In the case of Tom's pump, he presumably had reason enough to expect a higher discharge based on past experience. But it's not a fit case for simplistic generalisation.
On the same general topic of flow and head, the problem of emptying an overhead tank is probably too well known to be posed as a CR4 challenge. Will it empty faster if the water is discharged directly from the outlet, or when it flows through a pipe leading down to ground level?
To digress a bit (more), as is my wont, I make no claim of being a specialist or expert in turbomachines, but I've picked up some essentials of relating catalogue data to real situations, by having had to deal with a variety of pump/fan/blower installations.
I studied the subject long ago under a German professor whose knowledge and experience were indeed fantastic. Indeed I acquired much of my general design knowledge in between the lines of his lectures and by studying the diagrams of impeller profiles or pump components he drew on the blackboard, so effortlessly it seemed. Not many of my classmates took him seriously though, and some made fun of his accent. But I realised then itself that my mathematical limitations (coping with 3-D flow equations, vortex theory, boundary layers, ...) would prevent me from choosing that line as a career. I just stuck to simple general purpose mechanical engineering using conventional machine elements, and no regrets about that. If people today think my knowledge is outdated, that's their problem. I've immensely enjoyed my professional life, even though no one can accuse me of being a 'success', and I'm not too worried about my inability to communicate in computerese with the present generation.
4
