It is not unusual to pump product under vacuum. In oil refineries, there are distillation towers that run under vacuum to provide the final separation of lighter products out of asphalt. The pumps from these towers have slight vacuum at the suction. There are a few considerations. First, the pump has to have a positive NPSH (Net Positive Suction Head). Since this is dependant on vapor pressure, the fact that the suction line is at a pressure below atmospheric is not particularly relevant. If the product is at bubble point (boiling) at the surface of the liquid in the vessel, then the pump has an NPSH available that is equal to the elevation head up to the liquid level minus any losses in the piping. You should do a full NPSH calculation to determine how much head is available to the pump. You should compare this to the NPSH required by the pump. The NPSH available should be at least a few feet greater than the NPSH required.
Second, you need to be sure that the pump is operating at a good point on the curve. If the flow is too low (below minimum stable continuous) the pump will cavitate even if there is an adequate NPSH margin. If the flow is too high (>120% of Best Efficiency Point), then the NPSH required may have increased so much that it exceeds the NPSH available. If you want greater reliability, this range should be tightened further. We would attempt to keep this pump in a range between about 70% and 110% of BEP flow.
Third, you need to have positive pressure on the mechanical seals. If the seal runs under vacuum, it could “suck air” if it fails. In my business, a mixture of air and product inside a pump or pipe is a very dangerous thing. But, even if the seal does not fail, the negative pressure across the seal can affect the lubrication across the faces. We normally have our vacuum service pumps designed with restricted balance holes through the impeller to keep the seal chamber pressure positive. This has to be done carefully, by a qualified pump designer since it will also have a large affect on thrust loads. Positive pressure at the seal can also be accomplished with a seal flush plan such as API plan 54 using a separate pressurized stream between two seals.
The damage you describe does not seem to be to be tied to the fact that the pump runs at vacuum. It could be a simple mechanical problem. You should check the concentricity of the shaft within the bearing housing, the fit from the bearing housing to the head and the fit from the head to the case. If you mount the bearing housing on the head (with no seal) and sweep the concentricity of the seal chamber and the head wear ring, this will give you the result. Even if each of the fits is within specification, the tolerance stack-up could result in a situation where the shaft is not running down the center of the seal chamber or the head wear ring.
If the pump has two volutes, then you need to check that the condition of the volutes is good. The volute lips (cut waters) should be symmetrical and even. The volutes should be unrestricted (no foreign material obstruction). If the volutes were uneven, this would result in a side load on the impeller when running away from BEP flow.
The seal pilot to the head should be checked. The seal needs to be located positively to be concentric to the seal chamber. The seal installation needs to be performed correctly. The seal needs to be positioned on the shaft, but not tightened to the head until after the bearing housing is bolted solidly to the head. Then the seal should be bolted to the head and the drive collar locked to the shaft. This is especially important if this is a non-API seal that does not have a pilot fit to the head.
Johnny Pellin
