Staying Ahead of the Curve
A centrifugal pump operates at the intersection of its head-capacity (H-Q) curve and a system curve. The shape of a system curve is determined by hydraulic losses in a system (bends, restrictions, valve partial opening/closing, pipe-length-to-diameter-ratio), as well as the elevation difference between discharge fluid level and suction supply fluid level.
Thus, as system restrictions change (valves open or closes, pipes get plugged up), the system curve reflects these changes, forcing the pump to operate at the intersection of the same pump curve and a changed system curve.
It is desirable, of course, for a pump to operate most efficiently, at a best efficiency point (BEP) because it takes the least amount of energy. Also, operating at BEP makes a pump operate more reliably, because this is where internal loads, noise and vibrations are the lowest, as shown in the illustration below.
In fluid transfer applications, pipe friction is usually a predominant component of a system curve. Chemical plants are a good example of many such applications, from the hydraulics viewpoint.
In wastewater treatment applications, particularly lift stations, the other extreme is often prominent, i.e. for example, static head (lift) hydraulics, where friction losses are relatively small in relation to the static lift, as the incoming influent is “lifted” by the effluent pumps to the level of the main wastewater treatment facility. In such applications, the main change in a system curve is caused by a change in a wet well level that leads to the pump inlet, and the higher the well level, the lower the total static head.
A municipality in the Midwest has been experiencing problems with its influent pumps installed at the end of the wastewater collection tunnel, located almost 350 ft below the surface. These two 6,000-hp single-stage, double-suction pumps were plagued with reliability issues, such as impeller damage, wear ring failures, noise and vibrations.
Each pump typically operates at 400 to 600 rpm, and lifts up to 60,000 gpm (40,000 gpm is at BEP) from the bottom of the tunnel to the surface, some 350 ft up.
System head is mostly static, with friction head being almost negligible, which also denotes area of operation of a single pump, as well as two pumps operating in parallel mode. Curves are shown at various levels of the influent wet well. Most of the time, the level is such that a pump head is approximately 350 ft.
At 60,000 gpm (and 600 rpm), a pump requires approximately 35 ft of net positive suction head (NPSHR), which must be less than net positive suction head available (NPSHA), with some margin. (A 5-ft margin is typically considered good.) Disregarding some minor losses at the short run of the inlet piping and vapor pressure of water at ambient conditions, NPSHA is essentially equal to the water level above the suction centerline plus the atmospheric head (34 ft). Where there appears to be an interesting relationship between a pump head and the suction level:
H = 372 - hS = 338 - NPSHA
Thus, liquid level at the pump inlet determines the pump head. An interesting observation can be made regarding the narrowed boundaries within which a pump can operate due to such “linked” head and NPSHA. At low well levels, a concern would be to not run out of the adequate NPSHA. At high well levels, the pump head could become so low that its performance curve would force it to operate way to the right of the BEP, “running out” into the region where the pump would be experiencing high loads, instabilities, and—ironically enough—a lack of sufficient NPSHA (at such a high flow rate), despite the fact that the inlet well level is extremely high. The required NPSHR can still be higher, as it tends to “stonewall” at very high flow, deep past the BEP region.
Operating off-BEP. As discussed above, the pump is restricted to operation within a relatively narrow range of flows, centered on the BEP point. This restriction is due to NPSH considerations, but also due to efficiency reasons. At BEP, these 2,000-hp pumps have 90% efficiency, but it drops quickly to the left, and to the right, of the best efficiency point.
For example, at half flow, pump efficiency drops to 70%, and it shows similar loss of efficiency to the right of the BEP.
What does it cost to operate this type of pump per year? At 600 rpm and BEP (40,000 gpm), each pump consumes power as calculated from this basic formula:
BHP = Q x H x SG / (3,960 x eff) = 40,000 x 350 x 1.0 (3,960 x 0.90) = 3,928 hp, which is 2930 kW
If we assume a 24/365 operation, at $0.07 per KW x hr., the operating cost per year is:
2,930 x 24 x 365 x 0.07 = $1,796,676
If efficiency is reduced from 90 to 70% (20% difference), we can estimate the wasted (inefficient) power as:
2,930 x 20/90 = 651 hp (wasted), or, similarly, in dollars: 1,796,676 x 20/90 = $399,261 per year, per pump.
Operating at worn or enlarged clearances. Originally, these pumps were specified to have 0.035-in. diametric clearance between the impeller and casing wear rings. Unfortunately, stainless rings, also supplied by the manufacturer originally, did not leave much forgiveness to the heavy impeller weight, deflecting the shaft and resulting in rings contact. Once in contact, stainless rings galled, and sometimes seized, resulting in frequent repairs. To combat this, the users gradually opened the ring clearances, eventually reaching 0.100-in. diametral value.
This helped keep the rings from contact, but resulted in additional (hidden) losses, such as leakage from the high-pressure side at the impeller to a lower pressure side across the ring clearance. The increased flow leakage reached, as calculated, 3,500 gpm above the normal value, which is equivalent to another $140,000 in wasted energy per year, per pump.
Once this was recognized, the plant started to review options to change the rings’ materials to allow operation at the original design clearances.
One option being considered is composites, similar to what is routinely used by the U.S. Navy. Using composites, ring clearances can not only be reduced back to the original spec, but likely to an even smaller value.
As a more extended retrofit, a complete impeller, made from then composite material, is being considered as well. This would reduce the rotor weight by nearly 80%, reducing deflections, and thus minimizing a possibility of any contact between the rings altogether.