Impellers – why underfile?

Underfiling is the removal of material from the underside (also back side, “non-working” side, or suction side) of an impeller vane at the inlet or the exit.  In this article, underfiling will refer to the exit.  The terms “underfiling”, “back-filing”, “S2 cutting”, and “Sv cutting” are equivalent.  There are probably others.  A related term is “overfiling” which refers to material removal from the top-side or “working” side of the vane.  By the way, I’m not a stickler for proper punctuation – I prefer “underfile”, but purists are inclined to use “under-file” with a hyphen.

Pump manufacturers routinely underfile cast metal impellers.  Impeller underfiling is primarily a means of establishing dimensional uniformity at the vane exit tips.  Vane-to-vane uniformity, in turn, reduces variation in performance from one pump to the next.

Underfiling also “straightens” the vane outer diameter surface or “land”.  Conventional low production volume sand casting processes usually require a vane shape that tapers from a thicker section at the impeller hub side to a thinner section at the eye shroud side.  Additionally, any other vane-to-vane or vane thickness non-uniformity may show up in the finish machined land as an irregularity.  An underfile procedure straightens the vane lands to parallel edges.

An impeller with nicely shaped inlet and exit tips, cleanly ground cast surfaces and uniformly machined features has a certain aesthetic appeal.  It may or may not affect its performance, but it certainly affects emotional attitude.

Some manufacturers do not underfile, except when needed after a performance test.  High volume production impellers are often cast, fabricated, or molded with dimensional precision and the need for underfiling is eliminated.

Underfiling is also used to increase the pump head versus flow (H-Q) performance.  This can be thought of as either increasing head or increasing flow.  The potential for increasing head can be in the range of a few percent to as much as 10%.  Underfiling or overfiling may also bump up the pump efficiency a bit – usually not more than 0.5%.

The initial underfile process removes sufficient material to ensure that all vanes are underfiled uniformly.    Usually there is opportunity to remove additional material and achieve a new underfile condition.  This is useful if some additional H-Q performance is necessary after test or when the slope of the head rise-to-shutoff needs to be “flattened” through a combination of underfiling and trimming.  A variety of vane exit geometry modifications are used for performance adjustment.

Predicting the outcome of a given underfile can be accomplished analytically or by using historical test data and experience.  Some impeller designs will not respond well to underfiling in terms of increasing head.  CFD calculations may be warranted for high consequence applications.  A calculation procedure based on the Euler equation for head, an estimate of the impeller hydraulic “slip factor”, and details of the impeller underfile geometry can be used to bracket the upper and lower bounds of the expected change in head.  However, lacking prior experience with a given pump geometry, the analytical prediction should be considered as an informed engineering guess.  This is especially true when the underfile is accompanied by other hydraulic modifications.

In theory, impeller vanes could be underfiled to knife edge thickness.  This is not at all practical as the thin vane sections will quickly erode away or break off in service.  Also, thinning down the vanes to a small fraction of their original thickness may result in fatigue breakage at the impeller exit.  Apart from mechanical considerations there are diminishing returns in H-Q performance from thinning down vanes.  The practical limit of underfile thickness reduction should take into account the (head x density), the vane number and knowledge of the design’s susceptibility to fatigue failure for the given application.

The vane exit tips on high head impellers, such as found in boiler feed pumps and other high pressure services, may be filed to an elliptical tip shape to reduce stress concentrations.  Such an operation normally involves material removal from both top and under sides of the vane.  Careful attention is also given to the vane-to-shroud fillet radii.

Underfiling is a manual procedure.  It involves careful layout and skillful grinding. Templates are often used.  On engineered impellers, underfiling is usually performed by tradesmen possessing substantial experience in these procedures.

While not all impellers are underfiled, the procedure certainly plays an important role in the manufacturing of engineered-to-order or assembled-to-order industrial pumps.  The technologies in support of underfiling range from shop grinding to CFD engineering.

For an independent evaluation of pump impeller trimming and underfiling, contact an experienced consulting engineer who can help with your specific application.  See our Services page for an overview of engineering consulting services we offer.


  1. Mohamed Elgendy says

    Dear sir

    Thank you so much for such a good information ,it’s an honor to know you .

    Best regards

    • says

      You’re welcome Jon, and I appreciate the feedback. It’s been some time since I wrote the article, but it reminds me of just how much know-how and technology can go into the details. Best regards, Randal

  2. AIM says

    Your explaination gives a very clear understanding
    Will appreciate giving pros and cons of “Underfilling” and “Overfilling”

    • says

      I trust your New Year is off to a good start.
      The objectives of underfiling and overfiling are often different. For instance, underfiling will often measurably increase the head-flow output of the impeller. Overfiling may increase head and efficiency slightly but this is often scarcely measureable. Sometimes both underfiling and overfiling operations are done together to obtain vane uniformity, performance uniformity, control stress concentrations, or any combination of these. Underfiling alone is done much more often than overfiling alone. The specific design, manufacturing and performance objectives must be established in order to choose one method over the other.
      Yours very truly,

  3. Siddiq Ali says

    Thank you for enlightening us with a wonderful article sir.
    I was wondering if the under or overfiling has any affect on the flow obtained by the impeller.

    Thank you

    • says

      Yes. A change in pump total head from either overfile or underfile of the impeller will result in a changed pump head versus flow (H-Q) characteristic curve. Assuming the system characteristic remains unchanged, the intersection of the pump and the system will be at a new H-Q point. Let’s say the impeller is underfiled and the resulting pump H-Q curve is raised. The H-Q intersection point of pump and system will likewise be raised; both head and flow will be higher.
      Thank you for this comment – it’s a good one.

  4. Chet Bays says

    What tool do you use for the underfile?
    While visiting a pump company in Texas, I saw a carbide burr that was 1/4″ in dia and tapered down to 1/16″ with 2-3/4″ length of cut. I have contacted all the burr companies that I can think of and have been told that this would be a special order item. Does anyone build this as standard item?

  5. says

    Chet, pardon my jumping in… But maybe they get their tools custom made — a common practice in many industries. (Just Google ‘custom carbide burr tooling’ and you’ll see…)

  6. Ashim Das says

    Dear Randal,
    Thank you very much for your wonderful explanasion. I am a system engineer and purchasing Centrifugal pump. The pump specification says “The pump shall be capable of 5% head increase at rated conditions by replacement of the impeller with larger diameter”. One of the pump supplier offered us an impeller which is very close to the maximum impeller size. When i ask clarification about the 5% head increase the bidder reply is – “The impeller tips will be under-filed to meet the 5% head increase requirement”. My question is : should i accept the pump?

    • says

      Dear Ashim,
      As in many technical or life situations, there is no absolute ‘yes’ or ‘no’ answer. Having worked for a pump manufacturer for many years, I know that an impeller underfile option is occasionally offered to satisfy the 5% head increase requirement. As to whether this is acceptable for the particular pump offering you are evaluating, there are a number of technical issues to consider and taken together they are beyond the scope of this forum.
      Yours is a good question. If you’d like to have independent expert advice on this, please visit the Ekwestrel contact page and you can reach me by message or by phone.
      Yours very truly,

Leave a Reply

Your email address will not be published. Required fields are marked *