Hydraulic design capabilities respond to several key needs in pumping equipment. Oil and gas, chemical, desalination, power and other plants require an overwhelming number of pumps compared with other equipment. No matter the industry, operators need to expect high reliability and performance from pumps. Operations and maintenance teams are increasingly sensitive to cost control and reduction throughout the pump's life cycle. As markets fluctuate, plant performance must become more flexible to frequent production changes. As a result, facilities must quickly and reliably rerate pumps for different operating conditions. Hydraulic design is a strategic asset that can address these needs from different areas of the selected application.

Computational Fluid Dynamics

Traditional pump hydraulic design tools essentially rely on velocity triangles, which predict performance parameters—such as capacity and head efficiency—with a high degree of accuracy. The concept is relatively simple and takes little time, but consolidated traditional methods may limit the ability to capture all hydraulic characteristics. This results in poor performance predictions for conditions such as low specific speed conditions, high-energy pumps, drastic efficiency improvements and critical suction conditions, among others. Computational fluid dynamics (CFD) allows designers to predict hydraulic performances over the full spectrum of operation conditions. Differential equations describe fluid characteristics from a macroscopic level. The results of these equations show velocity and pressure values at any point of the computational domain. Once the entire pressure/velocity domain is calculated for a desired operating condition, the pressure/velocity data set must be processed to obtain the overall performance characteristics. Commercial codes are now designed and optimized to offer user-friendly tools for completing the entire calculation process. A CFD analysis would need the complete calculation domain. For pump applications, this includes a solid modeling of the entire pump and the operating conditions.

Cost-Effective Design

Computational resources were highly expensive a few years ago. Today, the process is more affordable and several pumps and their components—suction, impeller and discharge—are modeled using this method. This would allow the prediction of the complete pump performance without compromising computational resources. The reduced cost of hardware has widened the computational capability to analyze the complete pump configuration. Until now, solving complex systems variations required critical assumptions to be validated through time-consuming experiments. Several published results show how CFD analysis has correctly predicted pump performance without the need for further prototype testing. The full prediction of pump performance covers the entire performance curve, through cavitation simulations at the impeller up to shut off conditions. The scale of detail is thanks to the complete tridimensional representation necessary before completing any CFD simulation. Solid modeling tools are becoming standard among hydraulic designers. The tridimensional requirement represents a shift in traditional design. However, the immediate availability of pump models in their final configuration is worth a change in industry practice. These models open designers up to wider perspectives during the manufacturing process and its relations with the design phase.

3-D Printing

Following these developments in CFD methods, 3-D printing technologies have likewise left the laboratory and become a consolidated reality in many industrial applications. 3-D printing offers easier pattern development and faster, more economical casting for pumps. Different printing technologies offer reliable manufacturing alternatives even in case of complex geometries—especially when paired with CFD design methods. The combination of CFD and 3-D printing will eventually define a new path in the pump industry standard—a direct and fast process from design to casting and manufacturing.