Effects of different blade loading strategies applied to Low specific speed tube-axial fans

Abstract

Current industrial fans belonging to the tube-axial type offer the best performance for specific speeds higher than 6. For specific speed higher than 7.5, it is widely demonstrated that the performance requirements and efficiency of fans designed using the free-vortex blade load distribution can be matched, when not exceeded, by using other blade load distributions, also with advantage in terms of fan size reduction. On the other hand, the large number of free-vortex designs populating the vane-axial industrial fan type, suggests that the advantage associated with blade load distributions other than free-vortex reduces as specific speed decreases and vanishes at specific speeds lower than 6. This work presents and compares three fan designs conceived by the authors for specific speed approximately equal to 4.3. The fans feature hubto- tip ratio equal to 0.5 and differ to each other only in the blade load distribution that varies from free-vortex to rigidbody, passing through the constant-swirl one. The aims of the work are to quantify the pros and cons associated with the considered blade loading distributions, to identify the blade loading concept best suited to the range of specific speeds where hub-to-tip ratios equal to 0.5 are recommended for tube-axial fans and, to assess the capability of the design method originally suggested by the authors for fans with low hub-to-tip ratio. For this purpose, CFD-RANS analyses were carried out to preliminarily characterise all three proposed designs and identify the most efficient one. The most promising solution was manufactured by rapid prototyping, in the 315mm diameter size, and tested according to ISO 5801. The results demonstrate that the constant-swirl blade loading distribution offers good performance even at specific speeds close to the lower extreme of the usual tube-axial industrial fans application range and can be successfully designed using the improved version of the authors’ method presented here.