by Ron Astall, United Pumps Australia
In the last article we discussed the dangers associated with arbitrarily specifying pump running speeds. Based on specific speed theory, the recommendation was to aim for a pump with the best shape impeller and retain an open mind about the pump running speed.
For a given duty point, the Specific Speed (Ns) formula can be used to predict the likely impeller shape based on the expected running speed “N” and the head per stage “H”. If Ns is a low number, a less efficient pump can be expected. Last time we focused primarily on selecting pumps with a higher running speed to enable a larger Ns value and hence the possibility of a better impeller shape. Of course, we could obtain a similar result by selecting multistage pumps which would reduce the head per stage in the formula.
For single stage pumps, the assertion was made – controversially perhaps – that where a higher running speed provided for a better impeller shape, there was no reliability downside. In fact, I went as far as saying that in these circumstances, when comparing a low speed pump versus a high speed pump for the same duty conditions, the higher speed pump ought to be more reliable and will not have inherently higher wear rates than the comparable low speed pump. OK, the faster pump will be smaller and less expensive to purchase, but to many people, this is completely counterintuitive. How can a faster running pump be more reliable than a slow speed pump?
We must remember that we are comparing pumps operating in the same service conditions; i.e. the same flow and the same differential head. We are thus comparing a large slow speed pump containing a correspondingly large impeller with a more compact pump with a smaller impeller. The issues I am now going to consider are internal wear and erosion, bearing loads, noise and vibration and suction performance. To assist with this discussion, I have selected some 2980 rpm pumps and some 1450 rpm pumps for the same service conditions – See Figs 2 & 3 below.
In each comparison, the respective pump drawings are shown correctly scaled relative to each other. I have deliberately selected pumps operating a little below Best Efficiency Point (BEP) flow so that the effect of hydraulic radial side thrust can be seen. All pumps are double volute and the hydraulic radial side thrusts have been calculated on this reduced basis and are assumed to act in a direction around 60-70O from the cutwater.
The Resultant Radial Thrusts are the vector sums of the impeller weights and the hydraulic radial side thrusts. The Volute Outlet Velocity is the mean fluid velocity at the volute exits. For the Axial Thrust calculation, all impellers have been considered to be hydraulically balanced with equal diameter back and front rings and with balance holes through the impellers. For these overhung pumps, the axial thrust is thus simply the product of the inlet pressure times the effective seal balance area.
Wear is generally assumed to be related to impingement velocity to the power of 2 to 2.5, so the issue is primarily fluid velocity. To generate the same head at a lower running speed, a correspondingly larger impeller diameter is required, so in theory the fluid velocities in the volute should be pretty much the same in each case. In one of the comparisons, the volute outlet velocity is actually a bit higher for the low speed selection, but in general it is safe to say that casing fluid velocities will be the same and that there is no wear penalty at the higher running speed in the casing.
Where this may not hold true is in the suction side where impeller entry velocities will be typically higher; as also evidenced by the NPSHr values. In clean liquid service this is not an issue provided there is adequate inlet pressure. For slurry and solids handling; velocities must be kept low all round and for this reason low speed (and low head) pumps are preferred for solids handling.
For clean liquid service there is no wear penalty with a well selected higher rpm pump for the same service conditions
From Fig 4 below it can be seen that whilst a lower rotational speed will produce an inversely direct corresponding increase in bearing life, an increase in the bearing load will dramatically decrease bearing life to the power of 3 for ball bearings and to the power of 10/3 for roller bearings. The impact of the load is much more significant than rotational speed.
I have not bothered to calculate actual bearing life for the above examples, since in each case the pump designer will design the bearing frame accordingly. What is very clear however is that the bearing loads are much larger for the low speed alternatives.
The radial bearing loads are typically four to five times larger. This is due to the larger diameter and hence much heavier impellers that are required for the low speed selection. The larger peripheral area of these impellers will also impact the hydraulic side thrust at partial capacities.
Axial bearing loads in an overhung pump depend on the impeller hydraulic balance and the inlet pressure. The lower speed selection will normally have a larger shaft diameter, thus generating larger axial loads when there are significant inlet pressures.
Bearing loads are significantly higher for the slow speed alternatives, requiring larger bearing frames if the same bearing life is to be achieved.
Noise and vibration
For a given out of balance mass, a faster running speed will create a higher out of balance force and hence potentially higher vibration levels. To ensure acceptable vibration levels, a better balance grade is important for higher speed equipment.
As the lower speed pumps are doing the same hydraulic work but have a larger surface area it could be argued that there is potential for more radiated noise from the slower units, but I have personally no experience of this because pumpset noise levels are generally more dependent on the driver than on the pump. At higher speeds, acoustic driver fans may be necessary to ensure noise levels are not exceeded.
Lower speed pumps do have a theoretical advantage in terms of noise and vibration, however, from experience the potential for noise and vibration can be managed with suitable balancing and by managing driver noise levels. Historical test results for the above ranges of pumps show this to be the case; with acceptable noise and vibration performance at either running speed.
Modern hydraulic design techniques such as computational fluid dynamics and inducer technology can expand the application of high speed pumps in low NPSH service, but all things being equal, a low speed selection will have better (lower NPSHr) suction performance than a high speed unit in the same service.
If the suction conditions are challenging, this may preclude a higher rpm selection.
There is no doubt that a given pump will last longer when operated at slower running speeds. This is not what we are discussing here! We are comparing different speed pumps selected for the same hydraulic conditions.
We have seen that internal velocities, and hence wear, will be the same in single stage pumps selected for the same head. We have also seen that where a good high speed selection is available, a lower speed alternative will almost certainly be heavier, more expensive, and will have higher bearing loads.
The recommendation is to select your pumps on the basis of the hydraulic duty and specify your other primary requirements directly; such as NPSHA, noise and vibration. Then let the pump vendor select the best pump for the job; unrestrained by speed constraints. The best speed may be 2980 or 1450, or 980rpm or even lower. It depends on your hydraulic conditions.
If you end up with a faster running speed, pump reliability will not be compromised. Most likely it will be more economic, more energy efficient and more reliable.