In Lev Nelik’s May 2025 column (“Overfiling & Underfiling of Impeller Blades to Improve What?”), he discussed hydraulics design and the impact on performance as a result of underfiling and overfiling the impeller blades (or vanes). The following is an interaction between Nelik and longtime reader Lee Ruiz about the article.
FROM: LEE RUIZ
To: Lev Nelik
Hi Dr. Nelik,
In response to your May P&S article, I started with the expression for BHP = Q H sg / (3,960 eff).
Then, for the same flow rate and head, Q H, BHP2 eff2 = Q H / 3,960 = BHP1 eff1.
Rearranging, BHP2 = BHP1 (eff1/eff2).
If eff2 < eff1, then the ratio (eff1/eff2) is greater than 1 and BHP2 > BHP1.
If eff1 = 90% and BHP1 = 1,000-hp and (eff1 - eff2) = 1%, then
BHP2 = 1,000-hp [ 90% / (90% - 1%) ] = 1,011-hp.
The difference in horsepower (hp) is 1,011 - 1,000 = 11-hp.
This is approximately a 10-hp difference for the 1-point reduction in efficiency.
The power difference of 10 hp is equal to about 7.5 kilowatts (kW). So, using the formula, ∆$/yr = ($/kWh) (∆kW) (hr/day) (day/yr), ∆$/yr = ($0.10) (7.5 kW) (24 hr/day) (365 days/yr) = $6,535 approximately. I’m sure the cost per kilowatt hour (kWh) is much higher than the $0.10 that I used.
Thank you for your article on the subject of impeller exit vane filing. I haven’t seen much on this topic other than one type of filing to increase head and another type for improved efficiency. I don’t think I’ve seen any supporting reasoning or proof. I’m anticipating your follow-up article with in-depth informative details. In my filing analysis, I assigned subscripts “b” and “a” to the impeller with “filed” and “unfiled” exit vanes respectively.
For underfiling, the exit channel sectional area Ab is greater than the original area Aa since channel width hb > ha. Then, for continuity, exit velocity Vb is slightly less than Va for the same flow rate Q. Therefore, per the Bernoulli energy equation, exit pressure/head Hpb is slightly greater than Hpa similar to a diffuser effect. So, the pump Hb-Q curve is slightly elevated compared to the unfiled impeller Ha-Q curve.
Again, for underfiling, since Hb > Ha, more input horsepower is required, BHPb > BHPa. Therefore, my guess is that the filed effb-Q curve is slightly lowered since effb = Qb Hb sg / (3,960 BHPb).
For overfiling, the channel width hb = ha approximately (sectional area Ab = Ab). Then, using an exit velocity triangle, the meridional flow velocity for both filed and unfiled impellers is about the same, cmb = cma. However, the exit vane angle is βb < βa. Therefore, absolute flow velocity is cb < ca. This means that Hb = u cub is less than Ha = u cua. So, the pump Hb-Q curve is slightly lower than the unfiled Ha-Q curve.
Again, for overfiling, since discharge head is Hb < Ha, less input horsepower is required, BHPb < BHPa. Therefore, my guess is that the filed effb-Q curve is slightly higher since effb = Qb Hb sg / (3,960 BHPb).
I’ve seen some examples of impellers that have been both overfiled and underfiled. This may have been done to gain some efficiency while not losing head.
Also, any information that you could share on impeller entrance vane filing would also be of interest. Thank you for your articles.
Regards,
Lee Ruiz
Oceanside, California
FROM: LEV NELIK
To: Lee Ruiz
Thank you, Lee—very detailed and insightful analysis! Thorough as usual! And true, this sort of detailed hydraulic analysis of the impeller efficiency improvement techniques is not often presented, and so I think the readers of P&S will enjoy seeing your calculations and feedback.
Dr. Lev Nelik, PE
