4.1. Validation of the Numerical Model to Nominal Operating Conditions
To analyze the accuracy of the FBC model, the CFD simulation was developed from the data provided by the on-site measurements. By defining a longitudinal plane, xz
plane, located at medium height, it is possible to observe the static pressure and velocity contours, as represented in Figure 5
and Figure 6
The static pressure in the fan volute reaches very low values. In this zone, the dynamic pressure component is also important due to the high passing air velocity. In terms of static pressure field calculated, it is seen, as mentioned previously, that an average value of static pressure is set in the fan zone boundary.
Furthermore, it is understood that the fact that the air flow at the outlet is purely radial and tangential, makes that the zones with higher static pressures are located at the near-wall zones. This phenomenon creates “death zones” where the static pressure levels are lower, and around where swirl formation takes place. Finally, the pressure drops defined for the bag filter over z
direction, in the function of its length, can also be observed in Figure 5
shows the velocity field inside the AHU: (a) velocity contour and velocity vectors at medium height and (b) velocity streamlines inside the AHU.
Regarding the velocity contour (Figure 6
a), a maximum velocity of approximately 40 m/s was found. This velocity is reached in the fan volute, where there is a shrinkage of the air flow, as expected. The air flow at the fan outlet is purely radial and tangential, being in the near-wall zone where the air flow evolves. It is possible to verify that in the middle of the domain, the velocity is lower. If the fan real geometry was taking into account, due to the exit angle of the fan blades, the air velocity at the fan outlet would have an axial component.
The velocity streamlines presented in Figure 6
b show the high level of turbulence of the air flow in the study. The visible zones in grey are the fan wall and the bag filter frame.
The big issue in this study case is the swirl generated by the high levels of turbulence. By projecting the velocity vectors, tangentially, into a xz plane located in medium height, is possible to see the formation of a swirl. The swirl is a mechanism of energy dissipation, and, thus, a pressure drop source. This phenomenon is important and may have a big impact on the performance and energy consumption of the AHU.
To check the precision and applicability of the FBC model, the fan pressure increases and the air volume flow rate, are analyzed. As mentioned before, the considered fan static pressure increase is the difference between the pressure measured in the fan inlet and in the bag filter inlet. This is the differential pressure that is used to compare the experimental and CFD results. Table 5
presents an overview of the experimental and numerical results and its difference.
There are several conclusions that can be taken from Table 5
results. First, all the measurements were done by a five repetitions process per variable. Being the AHU in operation at the time of the experimental measurements, the values can vary considerably due to several uncontrolled factors. The air volume low rate error may be due to two different reasons. Initially, the fact that the static pressure considered at the inlet zone is the one measured at the fan inlet. Second, the experimental value for the air flow rate is not a measured value but an estimated value. As presented in Section 2.2
, the air volume flow rate is achieved from the fan differential pressure measured on-site, and the corresponding fan k factor. Beyond the uncertainties in the experimental measurements, the manufacturer approach to achieve the air flow rate worked by the fan has associated an error of 5 to 6%. Finally, uncertainties associated with the geometry simplifications done and the FBC model utilization have an impact on the air flow rate calculation.
When it comes to the fan differential pressure results, it is seen that the error obtained is 42.2%. There are some explanations for this, as the simplifications in terms of the fan volute geometry and the fan modeling approach used. Then, the installation method of the pressure tips inside the AHU becomes a problem. In this case, the pressure tips have, approximately, 50 mm in length. This allows the formation of micro-swirl phenomena around the tubes, adulterating the pressure values read. This problem does not affect only this parameter but all the others. Another cause for the error obtained is in the flow pattern at the inlet section. In the CFD simulation, the air flow at the inlet section is relatively stable and does not have any rotational component associated. In fact, the flow is relatively turbulent and highly rotational.
Another important parameter in the analysis is the fan pressure increase. As seen in Table 5
, an error of 21.5% is obtained for this parameter. It is relatively high, but there are some issues. The FBC model is very dependent on the quality of the fan curve used as input. In this case, using some tested working points given by the fan manufacturer a fan curve at nominal conditions is obtained. Although, in the fan datasheet, only four to six working points are given for nominal conditions. This leads to a fan curve that is not very reliable. Furthermore, to achieve the fan curve at real working conditions the similarity between fans approach is used, which is an approximation. All these factors combined have an impact on the results.
Looking at the results in the bag filter, the errors found in pressure are associated with the model used in the CFD simulations. As previously mentioned, for the bag filter modeling, using the on-site measurements, it is defined as a pressure drop in the function of its length (z direction). Thus, the pressure drops induced by the bag filter, in ANSYS Fluent software, is only in the Z direction. It does not happen in a real AHU, where the pressure drop in a bag filter acts in every direction. Still, due to the high level of turbulence of the air flow in circulation, this problem gains more meaningfulness.
Despite all the limitations of the FBC model, this model is validated and chosen to accomplish the CFD simulations that are presented in the next sub-sections. Its simple application, its low requirements in terms of computational effort, and hardware make this model very attractive.
4.2. Improvement of the Air Handling Unit Performance
After the evaluation of the applicability of the FBC model to model the air flow inside an AHU, the potential of the rectangular cuboid-shaped body downstream of the fan is evaluated. The installation of a cuboid-shaped body at the exit zone of the centrifugal proved to be an interesting option to decrease the turbulence levels downstream of the centrifugal fan [12
]. With this solution, most of the conduct area is covered, leaving only a small channel for the air at the near zone of the conduct walls to pass. The use of this body allows to increase the fan efficiency by reducing the vorticity downstream of the fan, transferring a substantial part of the kinetic energy into pressure energy (static pressure).
To this body, for this investigation, the name of Flow Control Unit (FCU) is given. Two different sizes of FCU are studied. The dimensions are 527 mm × 760 mm × 250 mm and 592 mm × 879 mm × 250 mm for FCU 1 and FCU 2, respectively. The mesh and boundary conditions used for these simulations are the same presented before.
shows the pressure contour inside the AHU with (a) FCU1 and (b) FCU 2. Represented as a grey box, in Figure 7
, the effect of the FCU 1 and FCU 2 can be observed. A general increase of static pressure in the near-wall zones was obtained due to the utilization of an FCU. When comparing FCU 1 and 2, the static pressure levels with FCU 2 are higher.
Regarding the velocity field inside the AHU, the FCU acts as a minimizer of the swirl generated in the near-wall zones and even lower with FCU 2. Figure 8
shows the velocity contour inside the AHU with (a) FCU1 and (b) FCU 2 it is clearly seen that the FCU reduces the vorticity downstream of the fan, by comparing it with the velocity vectors obtained in Figure 6
. The shrinkage between the FCU wall and the AHU wall provokes an increment in the air velocity, minimizing thus the swirl formation. By decreasing the air flow area of passage, the air velocity variation in the perpendicular direction of the AHU wall, x
direction, in this case, is increased. Therefore, the shear stress in the shrinkages increase. Despite the fact that the pressure drops increases in these zones, the decreasing of dissipated energy due to swirl formation is the most significant.
shows the static pressure curves obtained with and without FCU, under real conditions. At the fan exit, the static pressure obtained with FCU 2 is lower than the obtained without FCU, as in the case with FCU 1. This may be justified by the transformation of a part of the air flow energy into kinetic energy due to the shrinkages induced by FCU and, consequently, a decrease of the pressure energy (static pressure). In the transition between the fan outlet and the beginning of the FCU 2, there is an increase of the static pressure before stabilizing through the FCU 2. The reason behind it may be the decreasing of the kinetic energy due to the shocks of the air flow against the FCU 2. Being the cross-section area of the FCU 2 higher than FCU 1, these phenomena may start to have an impact and the static pressure rises. After the FCU, there is an air expansion and there are transformations of kinetic energy into pressure energy.
The inclusion of a FCU resulted in a static pressure rise improvement of 15.1% and 19.2% with FCU 1 and FCU 2, respectively.
However, the results obtained with FCU 2 can be misleading in terms of the possible real losses due to air flow shocks against the FCU 2 front wall.
The previous investigation allowed to ensure that the FCU has huge potential to produce a useful effect in the flow pattern downstream of a centrifugal fan. However, further improvement can be achieved by a geometry optimization of the FCU 1.
The influence of changing the downstream geometry of FCU 1 in the fan static pressure increase appears to be an interesting option to reduce the static pressure. Figure 10
shows the new design of the FCU inside the AHU: (a) view in perspective and (b) side view (right). As seen in Figure 10
, the modified geometry has conic features in its downstream part. This shape is similar to a common conduct transformation used to connect a rectangular section to a circular section. For this case, it is given the name of FCU Alpha to the FCU with modified geometry. The performance of FCU 1 and FCU Alpha is compared by assuming that their front section area (rectangular) and total depth are equal. In the case of FCU Alpha, the rectangular part has 100 mm depth, and the conic part has 150 mm depth.
The air flow pattern with FCU Alpha is presented in detail, with the contour at medium height and streamlines in Figure 11
. The velocity field inside the AHU with the FCU optimized is represented: (a) velocity contour and velocity 370 vectors at medium heigh and (b) velocity streamlines inside the AHU. Looking at the velocity vectors obtained with FCU Alpha (Figure 10
a), differences downstream are found when compared with the results obtained with FCU 1 (Figure 6
a). With FCU Alpha, the air flow develops more in the middle of the AHU, as can be observed at the bag filter zone. Although, no additional relevant differences are seen in terms of air flow patterns.
Apart from the velocity analysis, Figure 12
shows the static pressure curves obtained with FCU 1 and FCU Alpha, under real conditions.
Static pressure curves obtained with FCU 1 and FCU Alpha over the whole domain are presented. An improvement in static pressure rise of 6.1% with FCU Alpha relative to FCU 1 is obtained. The reason why this value is relatively low may be due to the small distance between the fan exit and the bag filter entry. The objectives with FCU Alpha and its downstream geometric features were to improve the stability of the air flow at the bag filter inlet and decrease the Reynolds shear stress at the boundaries and downstream of the FCU. By transforming a bigger part of the energy, when passing around the FCU Alpha, in pressure energy, the objective was to achieve a higher value of static pressure rise relative to FCU 1.
Despite the improvement in the static pressure rise with FCU Alpha, it is was not expected a lower outlet static pressure. Being static pressure improvement relatively low, this difference may be due to mesh problems.