Recently we were asked to balance a large ID Fan in the field. Our engineer took his vibration analysis readings on all the bearings and found what looked to be a large static unbalance in the rotor. He went through his normal balancing procedure and put on his weight where everything told him it should be. Unfortunately the vibration levels actually increased. Wondering if he put the weight in the wrong position, he went through his calculations and found everything was correct. Next he did a sanity check and started the fan without any changes to the fan to make sure the amplitude and phase repeated. The amplitudes were similar and the phase changed about 5%.
Since the numbers were close, although not exact, he was hesitant about what was going on. The customer insisted on trying to balance it again but we had the same exact problem. This time our engineer was sure he made all the correct moves and that something was not acting linear like a normal balance issue. So we recommended bringing the unit to our dynamic rotor balancing center so we could verify the balance condition. The customer was already two days late in starting up the plant and did not want any additional delays but saw no other options as the vibration was too high to continue to operate the machine.
We quickly saw the issue with the fan after a couple of runs on the balancing machine. The readings were always changing and we could hear material inside the fan. Our curiosity quickly had us to drill a hole in the area of the noise and suddenly all kinds of fine dirt and rust started to flow out from the opening. When it was done, we measured six pounds of material was taken from the center of the fan.
Now with all that weight removed, the fan balancing made sense once again. We made a precision balance on the unit to 2W/N and put it back into service. Before the unit started up some personnel took cover but it came up below 0.1 ips and running smoother than ever.
Our engineer learned to trust his instincts. When a rotor does not look like it is repeating, take another run or so to verify it and if the rotor does not show normal linear results then there is something else going on.