You brought up the good points of centerline discharge.
Loss of efficiency is minor, and actually there may be no loss of efficiency in many cases.
Speaking only about scroll pump with no diffuser:

The fluid in the scroll just before the discharge opening connection, is undergoing an energy transformation (with associated losses) from velocity energy to diffusion energy, in other words, high velocity fluid coming off the impeller is being slowed down.  The tangenital discharge actually allows this slow down to occur in the pipe manifold, so there is some loss of efficiency in the pipe system near the pump.

On a centerline pump the fluid is slowed down more inside the pump than a tangenital discharge, as the fluid turns the corners to get into the centerline discharge connection.

One quick way to look at it is this, the fluid has to be slowed down.  The tangenital discharge "cheats" by moving that energy conversion process (and associated losses) out of the pump into the discharge manifolding.

Therefore, when considering the pump AND the discharge manifold, there may be no difference at all in efficiency between centerline and tangenital.

PUMPDESIGNER

Interesting info, but I'm not sure I see it the same way.  Assuming the discharge pipes are the same size at the discharge nozzle of the pump (and the same for a tangential discharge and centerline discharge), the (average) velocity should be constant at that point, for the same flowrate.  The diffusion process (conversion of velocity to pressure) occurs in the volute with a typical 4- 8% "taper".

The directional flow changes, as that required for centerline discharge consumes energy (therefore a loss).  Any benefit of this loss as it relates to the diffusion process, I'm not sure.  Is this what your saying?

I tried to evaluate the difference while referring to Stepanoff and Kurassik, but all of the designs they show are tangential.  Centerline is predominantly a "recent" thing, evolved from ANSI B73.1m requirements.  And as far as I know, mainly a need to freely vent air w/o other means.

HomeMadeSin:
"Any benefit of this loss as it relates to the diffusion process, I'm not sure.  Is this what your saying?"
Yes, that is what I am saying.
Do you think that the results will vary depending on pipe size and high flow versus low flow?  In other words, will losses in high flow conditions be transferred from the pump into the pipe manifold on tangenital pumps?  I have seen many high flow tangenitals where the pipe is immediately blown up after the pump, i.e. 6" discharge connection on pump but 8"-12" discharge pipe is connected to the pump.

PUMPDESIGNER

I am sorry, PUMPDESIGNER.  Maybe I'm looking at it all wrong.  I hope you don't mind me "thinking out loud" here.  

For a given flowrate (say 1000 gpm), the average velocities would be the same at the plane of either pump's discharge (assuming same discharge size).  If an 8" pipe is mated to the pumps 6" discharge, the losses upon exit would (?) be a sudden enlargement and with the same velocities the losses would be (v1-v2)^2/2g - at least theoritically for either pump.

I must admit I'm a bit lost on this one.  I must not be thinking on the same level.

I think we are splitting hair's on this one regarding the efficiency differential between tangential or top centre-line discharge pumps related to the volute discharge configuration.

The cut-water / commencement of the discharge from the volute of either a tangential or a top centre line pumps would be the same assuming the same impeller /volute design, the only difference is the tang. discharge will have a straight expanding discharge branch as compared to the top centre-line which will have a curved expanding discharge branch. As I see it, but not being a theoretical pump designer I could well be far off the point, any difference in energy recovery (if any) would be inmeasurable for anything other than a purely academic study.
For me as a practical pump guy, give me the top centre-line  discharge pump everytime in preference to a tangential discharge pump. Less air entrapment,better distribution of pipe loads (which should be avoided anyway) easier to lay out and pipe upto etc etc.  

International College
Naresuan University
Phitsanulok
Thailand

Artisi,
Yep, we are talking small amounts, I think HomeMadeSin might agree with that also, but he will have to make that statement for I do not speak for him.

HomeMadeSin,
I am thinking out loud too, although I know that both of us could sit down and do a serious analyses I for one am not motivated to do so at this time, not interesting, probably because of what Artisi said.

Continuing discussion,
On tangenital pump with 8" pipe connected to the 6" pump discharge, fluid will leave the pump at high velocity and slow down in the 8" pipe, thus transferring the energy losses into the pipe system (losses occur when energy is transformed from velocity energy into head energy).

If I exaggerate the situation to clarify, if we have an open discharge with no pipe connected, the fluid leaves the pump at very high velocity but much energy is lost the moment the fluid clears the discharge of the pump.

Do you agree with some or all of what I just said?

PUMPDESIGNER

Thanks Artisi & PUMPDESIGNER.  I'm glad to hear that it is splitting hair, as the impression I had was a 2 - 9 point effect on efficiency depending on the transitions.  If I had thought it was such a small difference, I wouldn't have brought it up to begin with.  On the other hand, I do enjoy a good debate, regardless of the impact of the topic :)

Sometimes the questions / debates are of great interest and helps to keep our old brains active - especially when asked something we probably have never thought about or not thought about in a long time.
Keep them coming.
  

International College
Naresuan University
Phitsanulok
Thailand

From reading your comments and books on the subject, if I understood the subject rightly, for a given pump and specific speed, making the volute periphery less "tangential" and more "concentric" with the impeller axis would significantly reduce radial loads, in particular when pumping high specific gravity liquids, while the drop in efficiency would be low, say around 1%. Would any of the experts tell me if I am on the right track ?

25362 - I'm not sure I follow your comment on "concentric".  

A concentric volute can still be either tangential or centerline discharge, and is a different beast altogether.

And while a concentric volute does reduce the radial load, the efficiency difference is much more pronounced than 1%.

25362,
You are headed in the right direction with the radial loading concept.
I do not know an easy way to numerically state the issue exactly because there are too many variables including fluid mass, manufacturing differences, and the affect is heavily dependent upon the flow rate (above or below BEP).
The fluid travels further and turns a corner in a centerline discharge vs. tangenital pump,
Therefore, under some conditions at high flow rates around or above BEP, more fluid energy will be transformed from velocity into diffusion energy, resulting in a more stable environment around the impeller not unlike what a diffuser accomplishes.

PUMPDESIGNER

To HomeMadeSin, I didn't insinuate a totally radial volute, but one in which the last, say half, portion of it is radially concentric rather than expanding tangentially.

However, it seems this is not the subject at all. The location of the outlet nozzle is. And I'm sorry for not understanding the query and my straying away from the original topic.

   With today's practice of designing centrifugal pumps using Computational Fluid Dynamics (CFD) methods, there is no reason to believe that centerline casing discharge nozzles necessarily have to produce any lesser hydraulic efficiency than tangential discharge nozzles. Tangential (Archimedean spiral or scroll-type) casings are an historical throwback to the times when virtually all centrifugal pumps designs recovered impeller-induced velocity head by a gradual increase in flow area around the inside wall of the casing up to the volute tongue. Radial hydraulic thrust loads from circumferentially unequalized static pressures with this scheme probably led to the other diffusing concepts that later came into vogue including double volutes, vaned diffusers, vaneless diffusers and dump (no vanes) diffusion into either volute or concentric casings. The continued retention of tangential nozzles with these later non-tangential diffusion concepts is likely to represent a case of pump design inertia. The key to maximum centrifugal pump hydraulic efficiency is converting all the impeller velocity head into static head at the downstream pressure tap which is usually in the discharge piping not in the pump casing. At this point, the flow profile should be undistorted and without any rotational component of flow 1e, swirl. To help get this ideal condition a designer can even consider using an expanding area discharge nozzle or a short length of expanding discharge piping. Intra-pump-casing diffuser is generally critical for multi-stage pumps(where unrecovered static pressure cascades through the succeeding stages)and for jet engines where fast pressure recovery downstream of compressor blades is vital before the combustion process occurs to get best efficiency. For single stage water pumps, it matters not whether full static pressure recovery occurs within the pump casing or downstream prior to the head measurement probe site.
   Using CFD analysis that includes part of the discharge piping, the exit flow profile from the pump casing can be optimized for either a tangential or centerline discharge nozzle.

vanstoja,
I liked your analyses.
I agree that CFD will change things, but CFD will not change the existing pumps quickly due to cost.
It was a discussion in theory I suppose.

PUMPDESIGNER

To HomeMadeSin, I concur with you in that the subject of radial thrust may belong to another thread altogether.
However, just for the sake of exactitude...
Sam Yedidiah in his Centrifugal Pumps User's Guidebook: Problems and Solutions (Chapman and Hall), Chapter 27 titled "Modifying the Casing Geometry", when referring to remachining the part of the volute adjacent to the volute tongue to make part of the casing concentric with the impeller, clearly says, I quote: "Depending on the thickness of the casing walls, this procedure usually allows us to make about 5% to 20% of the volute perfectly concentric with the impeller axis. Such a change usually produces a significant reduction in the radial loads. While such a change in the volute usually reduces the efficiency, this reduction is rarely more than 0.25% to 1.5%. Unquote. Boldfacing is mine.

To PUMPDESIGNER, thanks for your comprehensive comments. The same book by Yedidiah in Chapter 10, titled "Axial and Radial Thrust and Balancing", considers ways to reduce radial thrust and even presents a graph of thrust as a function of flow rate, and a formula to estimate the (maximum) thrust at shut-off.

25362 - Thanks for the info, that's interesting.  I hadn't heard of Yedidiah.....I'll have to check out his book.

To address HomeMadeSin's original question from a pump users point of view, I have a population of over 1200 centrifugal chemical pumps (ISO/ANSI/and other non-standard designs)of either true concentric,tangential volute or centre discharge volute and I have found that:
1) Only the true concentric is freely self-venting.
2) Pipework designers/pipefitters/and mechanical fitters will always prefer to work on centre discharge systems due to ease of design and installation.
3) Never allow pipe systems to be supported by any end suction pump.
In the majority of cases any variation in efficiency or thrust forces etc between similarly sized tangential and centre discharge pumps is merely nominal. As far as I am concerned apart from our aged concentric pumps I have no design preference either way.

jet1749,
Good, very good comments.
I would give my left arm for any long term data you have maintained.
Perhaps you do realize how hard it is for many of us to obtain such data, almost impossible.
Seldom does anyone publish such things, perhaps they think it not valuable, but it is.

PUMPDESIGNER

PUMPDESIGNER, not sure I can help you with much in the way of the top flight data you seem to be looking for. Most of our official records are incomplete or missing (dating back to the mid 1950's) Most of our pumping equipment has been obselete since the 1970's such as Ingersoll/SIHI/Crane/Rhodes Brydon Youatt/SPP/LaBour/Worthington Simpson/Kestner/Mono,virtually all the newer stuff is either IDP/Flowserve/Durco/Warman. Our "unofficial" records are really maintenance records/reliability/overhaul information held on an access database, and pump performance data (curves) generated on our own test facility. We use the information generated from the access database to automatically identify the most unreliable pumps which we term out "top 20", then utilise RCA techniques to bottom out failures. As you can imagine most failures are due to veteran pumps poorly matched to inadequate systems. Sometimes we can correct the system, other times we have to re-engineer the pumps to allow them to cope (usual fixes are designing/fitting better L3/D4 ratio shafts and bigger bearings, manufacturing large taper bore seal housings fitted with flow modifiers, modular cartridge seals in place of component seals, fitting pump out vanes to impellers, materials upgrades, casting our own new/modified cases/impellers and bearing housings etc etc etc.) Most of the knowledge has been gained from good industry practice as applied to the design of modern pumps and then transfering this to troublesome vintage and veteran pumps. This has proved to be surprisingly effective as long as you know what you are trying to achieve and why, and also are aware of the correct technique to carry out the work. This has reduced pump failures from about 45 per month to around 13-15 per month. Due to the extensive nature of some of the modifications this has incidentally put me in the position of "pump manufacturer" according to the new european ATEX regs!

Jet1749,
Ahhh,the wonderful art of engineering,
Applying theory in novel situations,
Satisfying to the soul.

While I strive constantly to rid myself of some of the labor (and subsequent satisfaction) by systemetizing the knowledge to the point where I do not have to work so hard.

Good when possible I suppose.

PUMPDESIGNER

Thinking aloud it appears that since hot pumps (above 100^oC) are preferentially centerline supported to allow for symmetric thermal expansion of the wetted parts without excessively abrading the wearing rings, in what manner would a centerline discharge nozzle, and connections, affect free thermal expansion on hot pumps ?

Using quark's idea on adding post-scripts:

Believe it or not the human body has 656 muscles. A smile puts in commission 17, and a frown 43.