Influence of Bends or O bstruc tions a t th e F an Discharge O u tle t on the Perform ance of

C entrifugal FansBy L. S. MARKS,1 J. H. RAUB,2 a n d H. R. PRATT3

FSP-56-12

The form and d im en sion s o f th e in le t d u cts and in le t boxes o f a cen trifugal fan have been show n to have a pro­found effect upon th e fan perform ance. T he object o f th e investigations, described in th is paper, w as to determ ine w hether bends or obstructions close to th e discharge end o f th e fan have an y sim ilar in fluence. N o su ch influence was found. T he conclusion reached is th a t w ith a com ­plete fan housing o f good design , a bend or ob stru ction connected directly to th e fan d ischarge w ill have n o ap ­preciable effect on th e fan perform ance and w ill resu lt in th e sam e losses th a t would occur i f th e bend or obstruction were located a t a considerable d istan ce from th e fan .

TWO previous papers4 by one of the authors (with others) presented results of investigations conducted to deter­mine the influence of inlet boxes and inlet ducts on the

performance of a centrifugal fan. It was shown that the ca­pacity of the fan was greatly affected by the form and dimensions of the inlet structures. For constant rpm and with unrestricted discharge a poor inlet box reduced the capacity by as much as 60 per cent and a poor arrangement of bends in the inlet duct was found to reduce the capacity 40 per cent. The maximum efficiency of the fan, however, was affected only slightly by the form of the inlet box but in greater degree by poorly arranged inlet ducts. It was further shown that the capacity and ef­ficiency of the fan could be restored largely by the use of ap­propriate guide vanes in the bends in the inlet ducts.

It has been thought by many engineers that the performance of a centrifugal fan would be found to be similarly influenced by the form and dimensions of the discharge duct immediately adjacent to the discharge outlet of the fan housing and it was de­cided to investigate this matter. It would seem from a priori reasoning that any such influence would be slight. The struc­tures on the inlet side of the fan determine the velocity distribu­tion of the air at the fan inlet and consequently influence the fan operation. On the discharge side, the fan has already completed

1 Professor of Mechanical Engineering, H arvard University, Cam­bridge, Mass. Mem. A.S.M.E. Professor M arks was born in Bir­mingham, England. He received the degree of B.Sc. from the Uni­versity of London in 1892 and M .M .E. from Cornell University in 1894. He was with the Ames Iron Works, Oswego, N. Y. in 1894 and then went to H arvard University as instructor in mechanical engineer­ing. In 1900 he was made assistant professor and in 1909 was ad­vanced to his present position. Professor M arks is author of “ Steam Tables and Diagrams,” “Gas and Oil Engines,” “ Mechanical Engi­neers’ Handbook,” “ The Airplane Engine,” and has contributed numerous articles to the technical press.

2 Galesburg, Illinois. Jun. A.S.M.E. M r. R aub studied for six months a t the ficole Alsacienne, Paris, France, was graduated in 1926 with the degree of B.S. from Knox College, Galesburg, 111., and in 1929 received the degree of M.S. in Mechanical Engineering and Business Administration from H arvard University. After gradua­tion Mr. E aub was employed by the J. I. Case Threshing Machine Co. in the testing department, then by the Nash M otors Co. on pro­duction work and later was engaged by Ingersoll Steel and Disc Co. in the engineering department.

3 Draftsman, Federal Shipbuilding and Drydock Co., Kearney,

its work and the effect of bends or obstructions at the fan dis­charge presumably would be (1) to influence the conversion of velocity head to static pressure and (2) to increase the resistance against which the fan discharges in a way precisely similar to that offered by a more distant obstruction. The velocity dis­tribution of the air as it discharges from a fan is less uniform than it becomes after traversing a length of straight duct and, as the resistance offered by a bend or obstruction is proportional to the square of the velocity, the total resistance will be somewhat greater when the obstruction is located at the fan discharge. This difference, however, should be negligible. Apart from this, it would seem that the only effect of an obstruction on the dis­charge side would be its influence on the pressures and velocities of the approaching air.

Theoretical considerations show that the character of a fluid stream is affected by any obstruction which it approaches and that the influence of the obstruction extends upstream for an indefinitely great distance. The magnitude of this influence diminishes very rapidly as the distance from the obstruction in­creases and quickly becomes negligible. The disturbance is cal­culable for the simple condition of streamline flow of an ideal fluid of infinite cross-section. For this condition, with a spherical obstruction, the velocity of the approaching stream at a point two diameters upstream from the center of the sphere is dimin­ished by one per cent. With a cylindrical obstruction of infinite length with its axis normal to the stream, the velocity along the line two diameters upstream from the axis of the cylinder is diminished six per cent. With a flat plate of infinite length and for flow normal to the plate, the velocity at a distance of one and one-half times the width of the plate upstream from the center line of plate is diminished five per cent.

The magnitude of this disturbance is a maximum along the flow line approaching the center of a symmetrical obstruction and diminishes rapidly as the distance from the central-flow line in­creases.

For non-ideal fluids (having viscosity and compressibility)

N. J. M r. P ra tt was graduated in 1932 from th e W ebb Institu te of Naval Architecture and in 1934 received the degree of S.M. from H arvard University. He has had two years’ engine-room experience on both steam and m otor ships, served one summer as draftsm an with the Electric B oat Co., Groton, Conn., and is now employed as drafts­m an with the Federal Shipbuilding and Drydock Co.

4 “ Influence of In let Boxes on the Performance of Induced-D raft Fans,” by L. S. M arks and E . A. W inzenburger, A.S.M .E. Trans., vol. 54, 1932, paper FSP-54-16.

“Influence of Bends in In le t D ucts on the Performance of In ­duced-Draft Fans,” by L. S. M arks, John Lomax, and Randolph Ashton, A.S.M .E. Trans., vol. 55, 1933, paper FSP-55-9.

C ontributed by the Power Division for presentation a t the Annual Meeting, New York, N. Y., December 3 to 7 , 1 9 3 4 , of T h e A m e r i c a n S o c ie t y o p M e c h a n ic a l E n g i n e e r s .

Discussion of this paper should be addressed to the Secretary, A.S.M.E., 29 W est 39th Street, New York, N. Y., and will be accepted until January 10, 1935, for publication in a later issue of Transactions.

N o t e : Statem ents and opinions advanced in papers are to be understood as individual expressions of their authors, and no t those of the Society.

767

768 TRANSACTIONS OF THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS

O T H E R P I S C H D U C T S A T T A C H E O H E R ?

E l e v a t i o n V i e w

F i g . 1 A b h a n g e m e n t f o b T e s t s o n 3 8 - I n . S t u e t e v a n t F a n

and ..with^turbulent flow no adequate theory is available, but it would appear probable that the disturbance resulting from the presence of an obstruction would extend a shorter distance up­stream than with the streamline flow of an ideal fluid.

In the case of the discharge from the fan, there is turbulent flow and a limited cross-section of the steam. The obstruction (elbow, tee, etc.) will usually occupy the whole cross-section as seen downstream and such obstructions increase the total resis­tance against which the fan is operating but do not necessarily affect the performance of the fan proper.

T e s t A r r a n g e m e n t s

The investigations were carried out a t the Gordon McKay Laboratory of the Harvard Engineering School on the 38-in., double-inlet, radial-tip fan which had been used previously for the inlet investigations. The fan housing includes a short ex­panding portion on the discharge side. As the discharge condi­tions were to be varied, it was decided to measure air volumes on the inlet side. This is a departure from the methods of the A.S.H.&V.E. Standard Test Code but it is believed that the method actually employed permits an accuracy of measurement greater than is possible with the code. The inlet arrangements are shown in Fig. 1. The air enters through a well-rounded nozzle into a large circular duct. This is transformed into a rectangular duct which splits into two ducts connecting with the two inlet boxes. All changes in shape and dimension of the ducts, or of direction of the air currents, are gradual and guide vanes are located in the curved sections. The air enters the inlet boxes with flow lines which, it is believed, approximate closely to those obtained with the more usual inlet arrangements. The operating condi­tions were controlled by a grid a (Fig. 1) made up of 36 vertical wooden slats l l/i« in- wide and 38 in. high. The fan capacity was controlled by varying the number of slats in position. They were always spaced in such manner as to distribute the flow uni­formly over the cross-section of the duct.

The volume of air flowing was calculated from the static pres­sure at the middle of the parallel portion of the nozzle. This pressure was checked many times against the static pressure measured in the duct at 6 (Fig. 1), and in no case was any differ­

ence discernible. Similarly an impact tube located in the center line of the nozzle a few inches downstream from the nozzle and facing upstream always gave a reading of exactly atmospheric pressure. The nozzle coefficient was determined by traverse with a small impact tube, following the method of the Bureau of Standards (Research Paper No. 49), and a coefficient of 0.995 was obtained. I t is believed that the air measurements are accurate within one per cent.

The nozzle air measurements were compared wit h those ob­tained by pitot-tube traverses in the discharge duct following the Standard Test Code procedure. The fan housing and all joints in the ducts were gone over with great care to prevent leakage which in this case would be into the system. The only unavoid­able leakages were at the places where the fan shaft passes through the inlet boxes and, at these places, felt washers pressing lightly against the shaft were provided. The leakage must have been negligible. Comparative tests gave the pitot-tube volume measurements not exceeding two per cent and averaging less than one per cent greater than the nozzle measurement for fan capaci­ties between 30 and 100 per cent. As the pitot tube tends to read high under all circumstances, this difference may be re­garded as verification of the nozzle measurements.

The fan is shown in Fig. 2 and the housing in Fig. 3. In these tests the discharge bends, ducts, and all other obstructions were connected directly to the discharge outlet of the housing without any intermediate run of straight duct. On their discharge sides these structures were connected to straight runs of duct of length sufficient to permit the completion of regain of pressure and then discharged directly into the atmosphere. The straight runs of duct were about three diameters in length and their friction resis­tances have been neglected in calculating efficiencies so that these efficiencies are slightly low. In making comparisons with the condition of unobstructed discharge, the standard for compari­son had a straight discharge duct about three diameters in length and for this case also the friction resistance of the discharge duct was neglected.

The static pressure against which a fan operates is the differ­ence between the static pressure at the discharge and the total pressure at the inlet. The inlet in this case is at the entrance

FUELS AND STEAM POWER FSP-56-12 769

of the fan. Any change in fan performance resulting from the presence of these bends or obstructions may result from (a) the resistance to flow offered by the bend or obstruction and/or (&) the influence which the bend or obstruction exercises on the fan performance. When there is no perceptible change in fan performance as determined in this manner, both of these factors must be negligible.

The total resistance against which the fan operates is the sum of the static resistance and the velocity head. The velocity head was calculated from the mean velocity at the end of the dis­charge duct. This velocity is equal to the volume flowing per unit time, as determined by the inlet-nozzle measurement,

F i g . 3 F a n H o u s in g a n d I n l e t B o x e s F ig . 4 A r r a n g e m e n t of 9 0 - D e g B e n d s

to the inlet boxes. Traverses were made at this location with pitot tubes, following the procedure of the Standard Test Code. The total pressure was not constant across the sections, al­though the variation was slight. After investigating the total pressure distribution for various fan capacities, it was found that the average total pressure occurred always at certain loca­tions in the cross-section and in subsequent tests the pitot tubes were set in these locations for determining the total pressure at the inlet to the fan.

F i o . 5 A r r a n g e m e n t of T e e D u c t

On the discharge side a traverse of the cross-section near the discharge outlet of the fan housing shows considerable variation in static pressure—too great to permit the use of any observation in that location for determining the fan resistance. The static pressure at the final discharge of the air was always atmospheric pressure and this was taken as the static pressure against which the fan discharged. By this procedure the bends or other ob­structions on the discharge side of the fan were included as part

divided by the terminal cross-sectional area of the discharge duct.

T h e D is c h a r g e A r r a n g e m e n t s

Two commonly occurring arrangements of the discharge were selected for investigation; 90-deg bend and tee discharge ducts. The details of these structures are shown in Figs. 4 and 5, respec­tively. As the discharge outlet in the fan housing was square, it was possible to test with the 90-deg bend in two different

770 TRANSACTIONS OF THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS

orientations, discharging vertically upward and discharging laterally. Discharging vertically upward, the air passes through the bend with the same direction of rotation tha t it had when passing through the fan, while discharging laterally the direc­tion of rotation is changed to a plane at 90 deg to that in the fan. The tee discharge duct (Fig. 5) consists of a short duct 14 in. long, 36 in. square containing three butterfly dampers and connecting with a larger duct, 44 in. square, with a sudden en­largement. The dampers were always operated wide open and can be oriented either with horizontal or vertical axes. The large duct can be blocked at one end so that the air may be dis­charged at either one or at both outlets.

F i o . 6 P e r f o r m a n c e C o r v e s W i t h 9 0 - D e g B e n d s

To diminish the resistance of these structures, they were later provided with guide vanes which are indicated by dotted lines in Figs. 4 and 5. For the 90-deg bend, two guide vanes were used, concentric with the bend and dividing it into three channels of equal depth. For the tee duct an attem pt was made to divide the approaching air into ten streams of equal horizontal width. No attem pt was made, however, to control the sudden enlarge­ment of each stream.

R e s u l t s o f T e s t s W i t h a 9 0 - D e g B e n d

The performance curves shown in Fig. 6 are for three condi­tions: (1) With a straight discharge duct 9 ft 5 in. long; (2) with the 90-deg bend oriented horizontally and terminating in a straight run of duct 9 ft 5 in. long; and (3) with the 90 deg bend oriented as in (2), but fitted with two guide vanes, and terminat­ing in the same straight run of duct. For conditions (1) and (3) the curves are so close together that they may be considered identical; for condition (2) the total pressures and total ef­ficiencies at any capacity are less than for conditions (1) and (3), but the power requirements are the same in all cases.

Performance curves drawn for the same conditions as in Fig. 6 but with the 90-deg bend discharging vertically upward are identi­cal with those of Fig. 6. If the velocity of the air leaving the fan housing is uniform across the discharge section, there would be no reason to expect that the orientation of the 90-deg bend would have any effect on the performance of the combination of fan and bend. A vertical traverse by pitot tube near the discharge

orifice of the fan housing and in the median line is given in Fig. 7. This curve shows tha t the velocity is practically uniform over the whole cross-section thereby explaining why the orientation of the 90-deg bend has no influence on the performance of the combina­tion of fan and bend.

The identity of the test results for (1) a straight discharge duct and (2 ) a 90-deg bend provided with guide vanes indicates (a) that the resistance of the 90-deg bend is negligible and (b) that its presence does not affect perceptibly the action of the fan. The difference between the performance curves with and without the guide vanes must be ascribed entirely to the losses in the vaneless bend. This same loss would have occurred if the bend had been placed in a more remote location in the discharge duct. The performance of the fan itself is uninfluenced by connecting a 90-deg bend directly to the fan discharge.

R e s u l t s o p T e s t s W i t h T e e D is c h a r g e D u c t s

The variables in the operating conditions for these tests were: (1) The orientation of the dampers, either with vertical or with horizontal axes, (2) discharge through both branches of the tee or through only one branch, and (3 ) the use of guide vanes as shown by dotted line in Fig. 5.

F io . 7 V e l o c i t y - H e a d D i s t r i b u t i o n a t D i s c h a r g e F r o m F a n

The results obtained are as follows:The investigation of the influence of the orientation of the

dampers yielded entirely negative results. The performance of the fan was not observably affected. This result is interpreted as indicating that the flow through the dampers is substantially parallel to the axis of the duct.

Discharge through both branches of the tee duct is found to give better performance than through one branch only. This is quite marked when static pressures and static efficiencies alone are considered as shown in Fig. 8. I t is less marked, however, for

FUELS AND STEAM POWER FSP-56-12 771

total pressures and total efficiencies (Fig. 9) because of the doubled terminal velocity, for a given capacity, when discharging through one branch only.

The use of guide vanes has no appreciable influence either for single or for double discharge.

The static pressures and efficiencies with two-way discharge are found to coincide with the values for a short straight duct, but this result must be fortuitous since the terminal discharge areas are entirely different in the two cases.

F x o . 8 P e r f o r m a n c e C u r v e s W i t h T e e D u c t

(B ased on s ta t ic p ressu re .)

The total pressures and efficiencies are considerably less than for a straight-discharge duct. With the tee duct there is a sudden enlargement from a cross-section of 9 sq ft to one of 26.8 sq ft and, a t 40,000 cfm, the mean velocity changes suddenly from 74 to 25 ft per sec. The corresponding velocity heads are 1.23 and 0.14 in. of water, or a drop in velocity head of 1.09 in. The tests show that the loss in total pressure resulting from the use of the tee duct is approximately this amount whether guide vanes are used or not. The presence of guide vanes does not affect the magnitude of the sudden enlargement but might be ex­pected to guide the air so as to result in increased regain. This

F i g . 9 P e r f o r m a n c e C u r v e s W i t h T e e D u c t (B ased o n to ta l p re ssu re .)

result, if it actually occurred, was not of sufficient magnitude to be perceptible.

Additional tests were made to determine the resistance of the tee duct by connecting it to the end of a 60-ft run of straight duct. At a capacity of 40,000 cfm its resistance was in good agreement with the difference between the total pressures for the straight duct and the tee as shown in Fig. 9. I t would appear then that the only effect of the tee duct is to increase the resis­tance on the discharge side and that it has no influence on the fan performance.

C o n c l u s i o n s

I t may be concluded from this investigation that the opera­tion of the fan tested was not affected to any appreciable extent by sudden enlargement or change of direction of the air stream as it left the fan housing. I t is the opinion of the authors that the conclusions stated may be applied quite generally whenever the fan housing is sufficiently complete to give the discharged air a general direction of flow and a uniform distribution at the outlet. Any extension of the conclusion, however, to other types of fan and other arrangements of housing must be based largely on opinion.