LNG TRANSFER PUMPS: IMPROVING OPERATION

HIGH SUCTION PERFORMANCE PUMP INDUCERS CAN ELIMINATE CAVITATION

High suction performance pump inducers are critical to a number of industrial and aerospace applications where the required pump inlet pressure is near the vapor pressure of the liquid to be pumped. These include boiler feed pumps, aircraft fuel pumps, space vehicle propellant pumps and liquefied natural gas (LNG) transfer pumps.

Moving natural gas efficiently from production sites to where it is used, for instance, requires LNG transfer pumps that can drain more LNG from storage tanks. In space applications, high suction performance inducers can reduce launch costs through lower propellant tank weight, and increased propellant feed system stability and robustness.

The inducer’s role is to raise the pressure of the liquid to a sufficient level so that downstream components can pump the process liquid to the required pressurewithout cavitation. Cavitation occurs when the local pressure inside the pump drops below the fluid vapor pressure. At this point, a vapor cavity forms, usually on the blade suction surface just upstream of the inlet throat, which leads to three debilitating effects on a pump:

• Reduction in head rise of the entire pump when the vapor cavity approaches the size of the inducer inlet throat

• Blade damage from the collapse of the vapor cavity in higher pressure flows

• Transient instabilities in the pump from the interaction between the flow and the vapor cavity. These transients can lead to blade structural failure or can incite instability and failure in the pumping system and surrounding structures.

Critical factor

Inlet pressure is a critical factor in determining the likelihood of cavitation. An inlet pressure well above the vapor pressure is desirable to reduce or eliminate cavitation. However, in many cases, high inlet pressures are not practical.

For example, LNG transfer pumps operate at low pressures when the storage tank is nearly empty to extract as much LNG as possible. Space launch vehicles are weight limited, and higher tank pressure means more weight, which impacts launch costs. In these systems, it is desirable to keep the inlet pressure as close to the vapor pressure as the pumps will allow.

The net positive suction head (NPSH) is the parameter used to judge a pump’s capacity to pump without cavitation. This is typically defined as the inlet total head minus the vapor head of the propellant where the pump experiences a 3 to 5% loss in head rise. This point is used because it is a good balance between operating at the lowest possible inlet pressure, while still maintaining a safe distance from the total head breakdown point. In a typical pump, the head rise falls off rapidly, soon after the 3 to 5% head breakdown point. In addition, there are often large flow instabilities associated with head falloff above 5%, which can cause structural failures in high power applications.

Pumps able to operate at lower values of NPSH are more resistant to cavitation and have higher suction performance.The NPSH of a pump is dependent upon the impeller geometry, flow rate (Q), and rotational speed (N). To reduce the number of independent parameters and to compare different pumps, it is common to combine the NPSH, flow rate, and rotational speed into a non-dimension group called the suction specific speed (Nss) defined in Equation 1 below. Common usage in the United States is to use units of rpm, GPM, and feet, and to drop the acceleration of gravity constant which results in a dimensional parameter that is still used as if it were dimensionless. This parameter is often called NssUS and differs from the dimensionless parameter by a constant scale factor.

Figure 1 shows a plot of NssUS versus inlet flow coefficient for test data from a variety of pumps at their design point from high-flow coefficient to lowflow coefficient.

Demanding applications

High-flow coefficient pumps are common in many industrial applications and have NssUS values below about 20,000. Low-flow coefficient pumps are used in aerospace or demanding industrial applications that require very high suction performance with values of NssUS in the range of 25,000 to 65,000. One observation from Figure 1 is that the suction specific speed is inversely proportional to the flow coefficient.

At high-flow coefficient, the variation in suction specific speed with flow coefficient is somewhat flat, and suction performance changes are mostly seen through changes in the blade thickness and shape near the leading edge. Because of this, designers often fail to recognize the influence of design point flow coefficient on suction performance, and efforts to significantly raise the suction performance of high-flow coefficient pumps are often not successful. As the flow coefficient drops below approximately 0.20, the influence of flow coefficient on suction performance is much more pronounced.

A theoretical curve of suction specific speed versus flow coefficient is shown in Figure 1. It is a synthesis of two techniques that have been widely used, depending on the flow coefficient of the pump. At lowflow coefficients below 0.15, designers use Brumfield’s criteria to estimate an appropriate flow coefficient to achieve a required level of suction specific speed. Brumfield’s criteria are a reflection of the tradeoffs between the bulk inlet flow velocity, flow angle and blade tip speed on the suction specific speed.

The data in Figure 1 track the theory pretty well, with the variance probably due to specific blade shapes and thickness variations that play a role in the suction performance, but are not accounted for in the theory. At high-flow coefficients, the theory transitions into using an empirically determined blade cavitation coefficient that is dependent upon the shape and thickness of the blade near the leading edge. The match of test data to the theory in this region will be dependent upon the blade cavitation coefficient that is used.

Low-flow coefficients

Researchers have investigated inducer design at low-flow coefficients using theoretical analysis and water rig testing. Initial efforts focused on blade design methodologies for improving suction performance in the range of flow coefficients (0.06 to 0.12) needed for high-suction performance. Modest success was achieved in raising the suction-specific speed above the theoretical optimum shown in Figure 1. However, it became apparent that any significant increase in suction performance would require a reduction in flow coefficient. The problem is that there is a kind of stability barrier at a flow coefficient of approximately 0.06.

When approaching this point and below, it is difficult if not impossible to design an inducer that does not exhibit inlet backflow and cavitation-induced instabilities, especially if the inducer is required to operate at a low-flow, off-design point. As the flow coefficient decreases, the inlet relative flow angle as measured from the tangential direction decreases as well, and the blade angle must decrease with it. At low tangential blade angles, the inlet becomes highly sensitive to small changes in incidence angle, and the flow is more susceptible to stall and inlet backflow.

To overcome the stability barrier, researchers focused on inlet cover treatment devices to stabilize the flow at low flow coefficients. Several variants were built and tested, and the most successful incarnation improved the stability of lowflow coefficient inducers around the current stability barrier.

Efforts are ongoing to move the stability limit, shown in Figure 1, further to the left, using computational fluid dynamic (CFD) simulations.

Two-phase, time-dependent calculations were made on various configurations at different inlet pressures to simulate the cavitation head break down characteristics of an actual pump rig test. Figure 2 shows plots of vapor volume fraction contours in the computational domain at four inlet pressures. The normalized suction specific speed changes from 0.1 to 1.0, with the normalization factor as the suction specific speed at the 5% head breakdown point.

At 10% of maximum normalized Nss, there is significant vapor forming on the blades, which is typical of all inducers — cavitation inception always occurs well before head breakdown.

At the intermediate levels, there is a vapor cavity just upstreamof the inducer that is probably due to the interaction of the stability control device with the upstream flow. This upstream vapor cavity does not appear to have any impact on the inducer suction performance or stability. At the 5% head breakdown point, there is vapor cavity blockagewithin the inducer blades, as is typical for other inducers. In this case, it is running stably at a flow coefficient well below the current stability limit, and the head breakdown is at a much higher suction-specific speed than the current state of the art.

The next phase of development will be testing to gather information on head, efficiency, and cavitation performance, with particular attention paid to high-frequency pressure measurements to show the stability characteristics of the technology. After water rig testing, development efforts will focus on testing in applicable fluids, such as LNG and liquid hydrogen.

Author

Kerry Oliphant is Aerospace Program Manager/Corporate Fellow at Concepts NREC, a company that provides turbomachinery design, engineering services, manufacturing and CAE/CAM software. For more information, go to: www.ConceptsNREC.com