INTRODUCTION
Centrifugal pumps are flow machines that convert mechanical energy into pressure energy (Shankar et al., 2016) with a considerable energy utilization which can be the 20 % of the total consumption (Weidong et al., 2017; Lai et al., 2019). In recent years, pump manufacturers have intensified their quest to develop rapidly cost-effective and high-performance pumps with compact and robust structures to meet the unlimited consumers´ demands for high-end centrifugal pumps, as they offer a wide scope steady operation in both industry and agriculture (Wang et al., 2019).
According to Lai et al. (2019), the centrifugal pump internal flow is complex due to its three-dimensional and unsteady feature. The impeller rotation leads to periodic flow interference between the impeller and the vane diffuser. This periodic flow interference produces pressure and vortexes pulsations, which cause the main energy loss in the centrifugal pumps.
Traditionally, the information related to rotodynamic pumps behavior has been provided graphically, through the known characteristic curves of Load-Capacity, Efficiency-Capacity, Power-Capacity and NPSHr-Capacity (Aranibar, 2016; Iannetti et al., 2016; Martínez & Riaño, 2018).
Currently, Computational Fluid Dynamics (CFD) represents a common practice for designing and optimizing hydraulic pumps, as it can improve the pump design, while reducing development cost and speeding commercialization time (Lorusso et al., 2017).
Computational Fluid Dynamics (CFD) is one of the techniques used to solve Navier-Sotokes equations, which is rapidly gaining in importance thanks to the development of high-speed computers. The CFD method uses numerical approaches to solve the nonlinear differential equations that describe a fluid behavior under certain geometries and boundary conditions. The CFD main advantage is that it is a modeling technique, which allows engineers to evaluate a wide range of computer system configurations performance with less time consumption (Abo Elyamin et al., 2019).
Many numerical models can be found to calculate the Net Positive Suction Head NPSHr and evaluate the pump head curves. Generally, these models consider the complete headrace as the computational domain that shows great predictability of the NPSHr (Lorusso et al., 2017).
The foregoing shows that the CFD approach has been widely used in centrifugal pumps as a numerical simulation tool for performance prediction under design and off-design conditions, parametric study, cavitation analysis, analysis of interaction effects on different components, prediction of axial thrust, study of pump performance in turbine mode, analysis of diffuser pump, etc. (Shah et al., 2013).
In Cuba, there is a group of mechanical production companies that manufactures and repairs components to satisfy demands of important economic branches, such as the sugar industry. The commercialization of these pieces must compete with a dynamic market, which requires the development of new design projects with quality, speed and low cost. The need to adjust to current norms and standards is another factor, which regulates, as a fundamental parameter, the external dimensions and rotation frequencies. The power to start a pump is usually supplied by an electric or internal combustion motor, as long as it meets the specifications set forth by the manufacturer (Márquez, 2002).
The Industrial Technical Services Company ZETI: "Comandante Manuel Fajardo Rivero", located in Manzanillo City and belonging to the Business Group AZCUBA, which has traditionally worked in the manufacture of spare parts for the Cuban sugar industry and for export, has among its technologies foundry and is currently in an investment process. The investment plan includes foundry and machinery, which are the two main technologies of the company. To achieve its objectives, it has acquired CNC (Computational Numerical Control) machine tools with the appropriate software, as well as a high-tech laboratory equipment to guarantee the quality parameters of both technologies. In short, the company has increasingly moved its traditional technologies towards what is known as CAD / CAM technologies. Taking into account the abovementioned, the present work was aimed at simulate the performance of the BSC 80/200 (130-65) centrifugal pump using the CFD method, thus, obtaining the head-flow, power-flow and efficiency-flow characteristic curves.
MATERIALS AND METHODS
Operating Specifications
To analyze the pressures at the pump outlet, the impeller-volute assembly was studied. The geometric dimensions of each of the components correspond to the operating specifications shown in Table 1.
The design of the parts was initially obtained through the application termed Dipropump (Figure 1) created for the automation of the design (calculation and drawing) of the main parts of a model of a radial centrifugal pump of one stage and single suction, which is executed on AutoCAD.
The result was the technical documentation in two dimensions (2D) of the pump casing as well as the impeller (Figure 2).
The casing and impeller of the pump were modeled in the CAD SolidWorks program, and the dimensions were taken directly from the results of the execution of the above mentioned application (Figure 3).
Flow Field
The flow field depends on the type of analysis to be performed, which can be external or internal. In this investigation, the analysis was internal. In the Fluent-CFX calculation complement the liquid region was used as shown in Figure 4.
Fluid Characteristics
The design calculations took into account water as the fluid to be moved, which when passing through the interior of a centrifugal pump can experience phase changes from liquid to vapor and vice versa as a result of temperature and pressure action. Every fluid has a saturation vapor pressure for each temperature, water at 25 ° C has the properties shown in table 2 and a saturation pressure of 0, 03166 bar (Mataix, 1986).
Considerations for CFD Analysis
Computational Fluid Dynamics (CFD) is based on the use of numerical methods to solve the equations that describe mass conservation, momentum and energy of a fluid. There are several computer calculation programs for the numerical simulation of fluids, which require powerful computers not often available, so the calculation complement termed ANSYS CFX was used for this analysis (Yao et al., 2016; Domagała and Momeni, 2017).
Scheme for Simulation
To generate the simulation, the analysis systems and the system components were used in the Project Schematic; Figure 5 shows the scheme to achieve the analysis inside the pump.
Physical Definition of the Model
This interactive procedure was a pre-processing stage used to create the input required by the solver. The mesh files were loaded into the physics preprocessor, CFX-pre.
Once defined the domains, the boundary conditions simulating the different parts of the system (the mobile part, the fixed part and the interfaces between the different domains) were added (Figure 6).
RESULTS AND DISCUSSION
After carrying out the simulations for the three impellers (5, 6 and 7 blades), it was possible to corroborate the main operating characteristic of this type of hydraulic machine which consists in transforming the mechanical energy (from the motor) through the speed that the impeller communicates to the fluid (figure 7a), in hydraulic energy in the volute, translated into pressure energy at the outlet of the pump (Figure 7b).
The maximum values of fluid speed, approximately 35,2 m s-1,were found in the edge of the impeller and in the narrowest areas of the carcass, decreasing considerably in the discharge region. The pressures reached the top values in the area of the discharge flange, with 14 690 Pa, results that are close to those reported by Abo Elyamin et al. (2019) for the same number of blades in the impeller.
In Figure 8, the pressure behavior in the three variants of impellers analyzed considering the number of blades can be observed, showing that as the number of blades increased, the low pressure area was greater, which favored that cavitation appeared. Chakraborty & Pandey (2011); Chakraborty et al. (2013) and Abo Elyamin et al. (2019), when analyzing the effect of the number of blades on the performance of centrifugal pumps found similar results.
Figure 9 shows the impeller flow direction for the three cases under study, the speed variation caused by the number of blades can be observed. The fluid friction with the pump internal walls caused a decrease in the flow speed, which caused hydraulic losses. The simulation results showed that as the number of blades increased, the speed in the impeller inlet region and the blade decreased, increasing loss by friction. Therefore, the highest speed values in this region (impeller inlet and blade) were obtained for the impeller with 5 blades (Figure 9a), the mid values for 6 blades (Figure 9b), and the lowest values for 7 blades (Figure 9c). However, for this case (7 blades) there was the possibility that zones of turbulence and recirculation might appear near the tongue (Figure 9c).
The results emerging from the simulation coincide with those reported by Abo Elyamin et al. (2019), where the highest speed values were obtained for 5 blades and the lowest ones for 7 blades. However, for 9 blades, there was a loss increase in the areas near the tongue. Result that corresponds with those obtained in the present investigation for 7 blades and may have been conditioned by the fact that the rotation frequency was higher (3 480 min-1) than the one used by Abo Elyamin et al. (2019) which was 2,800 min-1. Chakraborty & Pandey (2011) also found this last tendency with the increase in the number of blades.
Subsequently, the results of the head-flow, power-flow and efficiency-flow characteristic curves were presented, for which the results of the pressure values equivalent to a range of 0 to 100% of the flow design were used.
As it can be seen in Figure 10, the correlation in all the curves was as expected, according to that reported in the literature (Pfeiderer, 1960; Church, 1987; Karassik et al., 2001) and the curve offering the highest values of discharge head was that corresponding to the 7 blades impeller.
As it can be seen in Figure 11, the impeller that required the highest power value is the one with 7 blades. The increase in the mass of this impeller justified its higher power consumption compared to the others studied.
Figure 12 shows the efficiency-flow characteristic curves for the three impellers analyzed. It can be said that the efficiency also showed a behavior proportional to the number of blades, since in the 5 blades impeller this value was over 80%, in 6 blades it was close to 90% and in 7 blades there were already efficiencies greater than 90%.
CONCLUSIONS
It was possible to obtain the three-dimensional models of the pump casing and the three impellers (5, 6 and 7 blades) with the help of the SolidWorks CAD program, from the 2D graphic information offered by the Dipropump application, which demonstrated the efficacy of its use.
The cases studied show the effect caused by the increase in the number of blades in the impeller on the different regions of the pump. It was obtained that with the increase in the number of blades the pressure increased gradually, the losses for friction increased and the fluid speed values decreased in the areas near the inlet of the impeller and blade.
The comparison of the results obtained through CFD simulations, taking into account the variations in the number of blades, showed that the impeller generating the best results was that of 7 blades, reaching better efficiency (90%) and power (36, 25 Kw) for a 70% of the flow design.