We were told that the 48-in diameter sewage pump impeller from one of New York City's main sewage pumps had been found "practically destroyed." When the impeller actually arrived at our location, both the shop foreman and the service manager asked, "Where is the other half?"

After realizing that there was no other half, they both looked puzzled and somewhat worried. Some of their concerns were voiced at our internal planning meeting, where we tried to set a proper course for the repair:

  • How are we going to build the suction half of the impeller from only an old "one line drawing" that really has no details of the vanes, especially the section that attaches to the suction ring?
  • How are we going to make sure that we do not considerably change the flow characteristics of the original impeller and risk doing a lot of work for nothing?
  • How are we going to "attach" the old half of the impeller to the new one, in a way that creates a sound and integral bond that will not fatigue and fail at a later date?
  • How are we going to measure/calculate the weight distribution during the positioning of the four new vanes in a way that it will not create a dynamic balancing nightmare later on?

Any of these potential problems would have been enough to discourage most repair facilities from undertaking such a challenging repair job. However, at the end of our strategy meeting the consensus was unanimous: we would apply our combined knowledge, field experience and basic common sense to develop and implement the best possible solution under these circumstances.

Several possible solutions were immediately explored:

  1. Get a new impeller from the manufacturer. This involved a waiting period of 48 to 52 weeks, a good long term solution, but one that meant having the pump out of service for one year.
  2. Call in several vendors so they could offer a more viable solution. Every vendor practically gave up, saying that since more than half of the impeller was missing, nothing could be done.
  3. Rebuild the impeller as close as possible to the original dimension. The customer provided us with a drawing of the impeller and we went to work.

The job was broken into several tasks and assigned to various employees to work on, as described in the sequences below.

Here is the piece left over of the original 48-in diameter impeller.

We had a foundry cast a ring and set it at the right height, following a customer's dimensional drawing.


Following the existing profile, we built a vane made out of reinforced plaster, then cut out the vane and sent it to a foundry to be reproduced.


The foundry provided four identical vanes. Here are two of them.

We then trimmed off the existing impeller vanes uniformly.


The four new vanes were carefully set in place and welded to the ring and to the hub.


The ID and OD of the suction ring was then machined according to the manufacturer's specifications.

Next, the impeller was put through an extensive phase of hand grinding to bring the four vanes to a uniform and dimensionally symmetric profile.

A stainless steel wear ring was then "shrunk on" and pinned to the impeller.

Next, the impeller was secured to a shaft provided by the customer.

Here are the impeller and shaft assembly being lifted and secured onto the balancing machine.

The operator gets ready to dynamically balance the rotor.

The entire job took only 5-1/2 weeks, allowing the customer to put a very important piece of equipment back on line at a fraction of the original cost estimate.

Pumps & Systems, September 2007