This action might not be possible to undo. Are you sure you want to continue?
13637 Received 12/01/2004 Accepted 15/07/2004 Keywords: floods & floodworks/hydraulics & hydrodynamics/models (physical) David E. Werth Assistant Professor, Department of Civil Engineering, Clemson University, Clemson, SC, USA Daniel E. Cheek Project Engineer, Hodges, Harbin, Newberry & Tribble Inc., Macon, GA, USA
Design guidelines for alternative formed suction inlets
D. E. Werth PhD, PE and D. E. Cheek MS
Formed suction inlets are often used to improve approach flow hydraulics to large vertical turbine pumps. Current design guidelines require that the pump bell be removed and the pump modified to allow for attachment of the formed suction inlet. The present research study was aimed at developing a dimensionless design procedure for a formed suction inlet based on pump bell diameter, which does not require removal of the bell, allowing for greater flexibility and economic feasibility for use in existing pump intakes. Model studies have shown this type of inlet to be successful at alleviating adverse hydraulic phenomena, but the results and design procedures are typically private and proprietary and are not readily available in the public domain. A formed suction inlet which can be constructed outside the sump and consists primarily of flat sides has been developed. The relatively simple geometry should minimise construction costs. In addition, the inlet is designed for use under existing pumps and does not require pump removal or modification. This inlet has been proved to be effective in a wide variety of pumping applications and is shown in Fig. 1. However, the relatively complex geometry can substantially increase the cost, and the need to remove the pump bell often limits its applicability as a corrective measure when retroﬁtting an existing pumping station. Therefore, it would be useful to have additional options and design guidance available for alternative formed inlets which are less costly and can be easily used on existing pumps. Private modelling laboratories have developed alternative formed inlet designs in the past; however, this information is often proprietary and not readily accessible for use by other design engineers. The present research is, in part, a result of numerous model studies which were conducted to develop alternative FSIs for a variety of inlet conﬁgurations. Each of the inlets was developed with the intention of utilising the existing pump bell. This paper presents the results of this study and proposes a set of functional design guidelines which can be used by engineers to develop an FSI based on pump bell diameter. 2. EXISTING KNOWLEDGE A commonly used type of FSI was developed by the United States Army Corps of Engineers (USACE) and is known as the Corps Type 10 inlet. The Type 10 design was originally published by Fletcher and Oswalt2 and entitled Geometry Limitations for the Formed Suction Intake. This document was recently rescinded from the public domain for unknown reasons. This design is also used as the standard for the Hydraulic Institute (HI) Standard Pump Intake Design manual.3 The Hydraulic Institute’s Standard suggests that the FSI may be a ‘ﬁx all’ for adverse sump pit hydraulics. However, the USACE Type 10 inlet is often considered costly and difﬁcult to build. The FSI has the potential to be very beneﬁcial for certain pumps and pump pit designs, especially for those pump sumps that are being retroﬁtted for a higher capacity or corrected for existing hydraulic problems. Antunes and Holman4 noted that the FSI has some tremendous advantages including its decreased sensitivity to unstable approach ﬂows, and the ability to raise sump ﬂoors because they require less submergence. This reduces the elevation of the impeller and the excavation required for the pump sump. This Werth Cheek 151
NOTATION BH inlet height at the back wall D pump throat diameter d pump bell diameter EH inlet height at the entrance IL overall inlet length IW inlet width at entrance W pump bay width
1. INTRODUCTION Formed suction inlet (FSI) devices have often been used on large vertical turbine pumps for a variety of reasons. They are relatively insensitive to high cross-ﬂow conditions, eliminate sub-surface vortex activity, and may reduce the required minimum pump submergence to minimise surface vortex activity. The authors have previously presented a summary of the advantages and disadvantages of FSIs and outlined the preliminary ﬁndings of the study.1 This paper is intended to expand upon the preliminary work and present the results of the experimental study.
The most commonly accepted, and only readily available performance and design guidance for this type of pump inlet has been from the US Army Corps of Engineers Type 10 inlet. Water Management 157 Issue WM3
Design guidelines for alternative formed suction inlets
Eight FSI designs were constructed and tested in four different inlet conﬁgurations during this study. The Type 10 inlet replaces the typical pump bell. the scale and ﬂow rate of sump conﬁguration 1 were varied in order to verify the inlet’s effectiveness over a range of ﬂow rates. with the exception of a simple radius at the back wall of the inlet. there are no standardised design criteria in which one set of dimensions may be applied to any situation. The ﬁrst ﬁve inlet Werth Cheek The overall length was measured from the entrance of the formed suction inlet to the back wall. Once an acceptable working design was found. not research studies. not the centreline of the impeller shaft. was referred to as sump conﬁguration 1 and was used to optimise a preliminary design that eliminated any undesirable hydraulic problems located inside the FSI. 2. it was assumed that the impeller diameter is typically 60% of the bell diameter. Tullis5 noted that for these reasons. There is a lack of information on FSIs in scientiﬁc literature. often these studies are typically applied to speciﬁc design criteria. Much of the work done to investigate FSIs is model studies of actual designs. 4. but is also economical and easy to build. Furthermore. The primary aim of the study was to develop an FSI which not 152 Water Management 157 Issue WM3 Design guidelines for alternative formed suction inlets . The main disadvantage of FSIs is that. The ﬁrst basin. not all of the examples reviewed in the literature were designed with back walls. and could be considered more of a concept or idea rather than a certain speciﬁc shape. As these Fig. In some cases. it was desirable to develop an inlet which could be constructed outside the sump. then lowered into the sump and placed directly under the existing pump bell. information regarding formed suction inlets rarely becomes available to the engineering community and results found during model studies are not shared among researchers.is especially helpful in areas with a high water table or when excavating in rock. In the past. 3. shown in Fig. researchers described inlet dimensions in terms of the impeller or throat diameters. DESIGN DEVELOPMENT To reach a starting point for the laboratory study. and could be implemented without emptying the sump. To use the Type 10 as a retroﬁt device requires removal of the existing pump bell and modiﬁcations to the pump itself. with the exception of the USACE Type 10 inlet. often with the aid of a physical hydraulic model study. Four separate basins.5–7 In order to directly compare the dimensions between FSIs based on bell diameter and those based on impeller diameter. each with unique approach ﬂow geometries. The range of values found during a review of literature for the overall FSI inlet dimensions was as follows: Overall Overall Overall Overall inlet width (IW): 1·39d–2·13d inlet length (IL): 2·05d–2·28d entrance height (EH): 0·53d–1·0d backwall height (BH): 0·24d–0·81d only provides uniform approach ﬂow conditions at the pump. a series of model tests were conducted with a variety of inlet geometries and approach ﬂow conditions. MODEL TESTING The major components used in the experiments were the model basins. the FSI and the model pump. cost is always a factor. This could be accomplished by developing an inlet which had primarily ﬂat panels. previous designs for FSIs were considered. were used in the design development. which eliminates the possibility of generalising the model study to develop substantial design data for the inlet itself. The FSI has little available documented information on its performance with reference to a particular design. 1. This has curbed its attractiveness as an acceptable alternative for solving hydraulic problems that occur in pump sumps. This would eliminate the need to remove the pump or pump bell. This is not always possible in existing stations where it may not be feasible or ideal to shut down the pump for an extended period of time. USACE Type 10 inlet studies are usually part of the design process. researchers used this formed inlet idea and applied it to their speciﬁc criteria. To accomplish these aims. A secondary aim was to develop an inlet which could easily be used to retroﬁt an existing pumping station. Therefore. However.
Some designs were tested to determine the optimum dimension to both minimise size yet provide acceptable conditions. The studies were conducted according to the 1998 Hydraulic Institute Standards. Sump configuration 1 conﬁgurations were used to develop an acceptable working design based on dimensionless parameters. In particular. curved backwall. velocity and turbulence variations were less than 10% and no vortices greater than a type 1 or weak type 2 were observed entering the inlet. Inlets that failed to meet the established acceptance criteria were modiﬁed until acceptable. These measurements included vortex activity. The last three were built using these parameters and tested in different approach ﬂow conditions. A summary of the different designs is shown in Table 1 and more complete details can be found in a master’s thesis by Cheek. Each model was constructed as an undistorted Froude scaled model with a length scale sufﬁciently large to ensure that the Reynolds number (Re) at the model pump bell exceeded 16105. Figs 3 and 4 show two of the model inlets To evaluate the effectiveness of each of the designs. pre-swirl of ﬂow entering the pump. turbulence levels. it would still be far less intense than that required to ingest or pull air out of solution. The inlet was deemed acceptable if pre-swirl was less than 58. A summary of the 0˝5d 2˝42d prototype pump and Froude 0˝92d 2˝42d 0˝75d 2d scale model information for 0˝75d 2˝5d each sump conﬁguration is 0˝75d 2˝25d shown in Table 2. Inlet design 1 2 3 4 5 6 7 8 Width 2˝78d 2˝78d 2d 2d 2d 2d 2d 2d Backwall height 0˝5d 0˝5d 0˝5d 0˝5d 0˝5d 0˝5d 0˝5d 0˝5d Table 1. was constructed based on the one principle that gives the Werth Cheek 153 Water Management 157 Issue WM3 Design guidelines for alternative formed suction inlets Flow to lab pump Trash screens to be located here (typ) . and turbulence levels within the pump.Flow straightening baffle FSI Control valve (typ) Orifice flow meter Baffle wall Circulating water pump (typ of 3) Control valve Flow Piers (typ) Elbow Flow from lab pump Plan view False walls Fig. and vortex activity are evaluated. such as a well-developed type 2 or very weak type 3. 2. It should be noted that while care was taken to minimise scale effects by ensuring fully turbulent ﬂow in the model. Summary of model configurations As the inlets were designed using models of existing intake structures. vortex formation may be slightly less intense in the model than in the prototype. RESULTS 0˝75d 2˝5d The ﬁrst FSI. pre-swirl.3 The HI Standards indicate several acceptance criteria that are to be used when evaluating this type of structure. velocity distributions around the throat of the pump. with the exception of a simple-radius. Should the vortex be slightly stronger in the prototype. the inlets themselves were constructed at the Entrance height Overall length same scale as the intake structures. over 60 tests were conducted at varying ﬂow rates and water levels. a series of measurements and observations were recorded during the testing of each inlet.8 Each of the designs tested was constructed entirely of ﬂat panels. particularly regarding air entrainment. inlet design 1. 0˝75d 2˝5d 0˝75d 2˝5d 5. velocity distributions. In all. To overcome this limitation. no vortex greater than a weak type 2 was permitted in the model.
Table 3 presents a summary of the results of the ﬁnal inlet 4 design when placed in sump conﬁguration 1. This causes the ﬂow to accelerate. and centre ﬂoor splitter. At the minimum recommended water level. However. To verify the need for the vertical wall at the entrance to the inlet. 2 FSI a distinct advantage over conventional wet-pit intakes: a constantly decreasing cross-sectional area. Type 3 surface vortices were harder to ﬁnd and broke up quickly. two sidewall ﬁllets. the 2·78d width of inlet 2 was less than ideal for retro-ﬁt applications. test inlet No. Inlet design 2 was then tested in sump conﬁguration 1 and was found to still have some ﬂow separation along the ﬂat part of the roof at the entrance where the turning vanes were located. a vertical curtain wall was extended from the front of the inlet to above the water surface. After testing this design in sump Fig. Inlet design 2 was built exactly as inlet design 1 with two exceptions. Although inlet design 2 met the acceptance criteria for vortex formation most bay widths for pump sump pits are only 2d. Instead of the entrance height being equal to the backwall height.conﬁguration 1. tests were conducted with and without the wall in place. and a vertical backwall extending from the backwall ﬁllet vertically to the top of the inlet. which eliminated the separation at the turning vanes. In addition. The initial geometry was chosen to ﬁt within the pump bay of sump conﬁguration 1. it was increased to 0·92d. 3. The water level was chosen as the minimum suggested level as indicated in the 1998 HI Standards. Four turning vanes were also added at the entrance to straighten the ﬂow entering the inlet. Next. Inlet 4 performed very well and was well within the established criteria. This eliminated surface vortices and prevented them from entering the pump. test inlet No. 4 154 Water Management 157 Issue WM3 Design guidelines for alternative formed suction inlets . The turning vane conﬁguration and entrance were modiﬁed to eliminate the need for a radius at the top of the inlet. it was noted that type 4 surface vortices were entering the inlet and there was a large separation zone occurring along the roof of the formed inlet. Inlet designs 3 and 4 were modiﬁed to reﬂect the bay widths and bell clearances recommended in the 1998 HI standards. which helps to eliminate vortices and dampen adverse ﬂows caused by poor approach angles. a 0·3d half-round piece was added to the top front edge of the entrance. It had an internal geometry consisting of a backwall ﬁllet. the pre-swirl angle and velocity data were within criteria. Final formed suction inlet. but were observed entering the inlet. Formed suction inlet. type 2 surface Werth Cheek Fig. 4.
0. 6. It was also noted that at water levels 10% below the recommended minimum submergence. Q: m3/h [gallons/min.33d IW = 2. Several points around the pump bell exceeded the Werth Cheek 155 Water Management 157 Issue WM3 Design guidelines for alternative formed suction inlets . These inlets were constructed based on the bell diameters of pumps for three uniquely different sump conﬁgurations. However. ADDITIONAL VALIDATION To further validate the ﬁnal design. including one with signiﬁcant cross-ﬂow. Summary of inlet 4 results vortices were observed with and without the wall in place.56d 0.] 52˝6 52˝6 49 49 42˝9 42˝9 40˝9 40˝9 [20˝7] [20˝7] [19˝3] [19˝3] [16˝9] [16˝9] [16˝1] [16˝1] Pre-swirl: degrees 1˝2 1˝2 0˝8 1˝3 1˝5 1˝0 1˝0 1˝6 Surface vortex intensity Type Type Type Type Type Type Type Type 2 2 2 2 2 2 2 2 Sub-surface vortex intensity Type Type Type Type Type Type Type Type 1^2 1^2 1^2 1^2 1^2 1^2 1^2 1^2 Vertical wall No Yes No Yes No Yes No Yes Velocity criteria met Yes Yes Yes Yes Yes Yes Yes Yes Turbulence criteria met Yes Yes Yes Yes Yes Yes Yes Yes Table 3.15D B A Turning vane 0.] 289˝6 236˝2 193˝0 121˝9 91˝4 162˝5 188˝0 116˝8         Model scale 15˝7 12˝8 10˝5 6˝6 5˝0 9˝25 9˝45 7˝26 Model flow. An additional test was conducted with inlet conﬁguration 5 which shortened the inlet slightly to 2·25d. The additional sumps were referred to as sump conﬁgurations 2. Again. 5. The design development testing led to a relatively simple FSI which can be easily constructed outside an existing sump and installed by divers.42d Backwall fillet Vertical fillet Floor splitter 90° A Fig. while vortex intensity was unchanged with the wall in place. it is recommended that the vertical wall be included. the addition of a vertical wall greatly reduced the sensitivity of surface formation to water levels. Model parameters Prototype flow rate: m3/h [gallons/min.Sump configuration 1 1 1 1 1 2 3 4 Prototype bell diameter: cm [in. The dimensionless inlet design is shown in Figs 5 to 7. without dewatering the pumping station. and with minimum station down-time. 7 and 8) were constructed based on the dimensionless parameters developed during inlet 4 design. 9 and 10 respectively. Visual observations indicated that the ﬂow was less uniform within the inlet and that the slightly reduced length was not beneﬁcial. Due to the sensitivity of surface vortex formation to water level. Actual model studies for these sumps were conducted and the ﬁnal formed inlet was installed and tested. Q: l/s [ft3/s] 14˝6 16˝1 17˝7 22˝4 23˝2 14˝6 17˝4 13˝3 [0˝52] [0˝57] [0˝63] [0˝79] [0˝82] [0˝51] [0˝61] [0˝47] Prototype flow. Preswirl and turbulence levels were well within criteria. much stronger air-entraining surface vortices were observed when the wall was removed. 3 and 4 and B are shown in Figs 8. Plan view of the recommended design Sump conﬁguration 2 was tested for a variety of ﬂow rates. these vortices tended to dissipate and break up as the entered the inlet.] 51 330 [226 000] 34 068 [150 000] 22 712 [100 000] 9084 [40 000] 4996 [22 000] 13 627 [60 000] 17 170 [75 600] 6814 [30 000] Reynolds number (at bell entrance) 1˝0 ¾ 105 1˝1 ¾ 105 1˝2 ¾ 105 1˝6 ¾ 105 1˝6 ¾ 105 1˝1 ¾ 105 1˝1 ¾ 105 1˝1 ¾ 105 Table 2.0d Sidewall fillet ∆d 0. three additional inlets (inlets 6.] 4996 [22 000] 4996 [22 000] 9084 [40 000] 9084 [40 000] 22 712 [100 000] 22 712 [100 000] 34 068 [150 000] 34 068 [150 000] Model scale 5˝0 5˝0 6˝6 6˝6 10˝5 10˝5 12˝8 12˝8 Model water level: cm [in.
eliminating the need for a dividing bay wall between inlets. but it was later approach ﬂow conditions. The introduction of the two Fig. a vertical 2. 7. Fig. 8. Velocity. Although these tests demonstrated determined that a misalignment of the pump bell above the inlet was causing some ﬂow separation at the pump bell.75d and required the use of a Floor splitter BH = 0.42d vertical vanes rather than three at the entrance resulted in less ﬂow separation at the inlet entrance. Control valve . which was nearly twice the HI recommended maximum cross-ﬂow of 0.55d straight approach ﬂow. effectively Vertical fillet Turning vane simulating an intake without dividing bay walls. pre-swirl and turbulence levels were well within the established criteria and overall conditions were extremely uniform within the sump.22d 25% of the bell velocity.0d ﬁllet that was installed near the back of the pump improved Section A conditions during high cross-ﬂow events. will be recommended regardless of the maximum allowable velocity deviation. Sump conﬁguration 4 was tested to investigate the impact of cross-ﬂow directly in front of the inlet. A prototype cross-ﬂow of Sidewall fillet 0·78 m/s (2·5 ft/s) was present in front of the ﬁrst inlet.) apart to Backwall fillet facilitate a vertical pier which was used to attach the vertical wall above the inlet entrance. End view of the recommended design additional inlet vanes as well as the vertical backwall ﬁllet are relatively minor modiﬁcations from the design used with straight approach ﬂow conditions and rather than have two different conﬁgurations. some structural support will probably be required to support the vertical wall and a short bay wall or pier could be placed between Floor splitter inlets to provide this support. Elevation view of the recommended design However. IL = 2 5d and inlet is 1 25 cm (0 5 in. This conBell centred above hole in top of inlet ﬁguration was not effective Sidewall fillet EH = 0.5d Backwall fillet 0.3d 0.5d vertical wall above the inlet 45˚ similar to the conﬁgurations 0. which agrees with previous modelling experience. The inlets in sump conﬁguration Vertical fillet 4 were placed a prototype distance of 30 cm (12 in.Sump conﬁguration 3 was tested to investigate the impact of a sloping ﬂoor upstream of the inlet. 6. . It was found that a series of ﬁve 0. The Note: Inlet is bolted to floor . Velocity. 1 75d Max clearance between bell vertical wall prevents ﬂow . In addition.) from travelling past the Section B entrance to the inlet. Tests were conducted without Vertical wall a vertical wall. Sump configuration 2 Design guidelines for alternative formed suction inlets Werth Cheek To lab pump uniform within the intake. pre-swirl and turbulence levels were well within the established criteria and overall conditions were extremely 156 Water Management 157 Issue WM3 Flow Flow Sluice gate (full width) Screen chamber Sluice gate (full width) Screen chamber Baffle wall Plan view Fig. .25d that were tested with a 1. This indicates that proper From lab pump alignment of the bell over the opening in the inlet is essential.
6 and 7 illustrate these dimensions.Control valve Orifice flow meter From lab pump To lab pump Fig. The inlet is bolted or ﬁxed directly to the ﬂoor beneath the pump bell. distance from back wall to where the inlet should begin to rise up to the entrance height. A FSI design which is based on the pump bell diameter is proposed as a viable alternative for eliminating adverse ﬂow phenomena occurring in existing wet pit pump sumps. WERTH D. 0·75d. Werth Cheek 157 Design guidelines for alternative formed suction inlets . the following conclusions and recommendations can be made. as demonstrated with sump conﬁguration 4. in turn. could result in tremendous savings. d. (d ) This formed suction inlet design was model tested for ﬂows ranging from 4996 m3/h (22 000 gallons/min) (prototype ﬂow) to 51 330 m3/h (226 000 gallons/min). Flow (b) This formed suction inlet design was not affected by severe cross-ﬂow conditions. Sump configuration 4 the effectiveness with cross-ﬂow velocities of nearly twice the HI recommended value. 10. and the recommended design is based on water levels equal to or greater than suggested by the 1998 HI Standards. and sidewall ﬁllets should be included. 1·5d. back wall height. Although it may be possible to signiﬁcantly reduce the water level to below the minimum submergence suggested in the 1998 HI Standards. 2003. Finally. 3. Proceedings of FEDSM’03 4th ASME–JSME Joint Fluids Engineering Conference. (a) The optimum overall design dimensions for this FSI design. as well as the laboratory experiments conducted in a controlled environment. Water Management 157 Issue WM3 backwall ﬁllet. so that it will be easily manufactured and assembled. 5. and CHEEK D. Sump configuration 3 Flow Influent pipes Vertical wall Fig. Furthermore. Hawaii. using ﬂat components. An alternate formed suction inlet design for large vertical turbine pumps. 0·5d. The ﬁnal design is applicable over a range of ﬂow rates and approach ﬂow conditions and effectively meets the acceptance criteria mentioned previously. resulting in minimum down-time for the application the pump is serving. Furthermore. Consideration of a model study should be made when applying this design to ﬂows out of this range or in conﬁgurations that may not be representative of those investigated during this study. entrance height. CONCLUSIONS The purpose of this paper is to present an efﬁcient and economical FSI design that meets all of the 1998 HI acceptance criteria. The material used to construct this FSI will most likely be concrete or steel but should be speciﬁed by the design engineer for the speciﬁc application. further research is required to determine the upper limits of this value. The ability of the FSI to straighten the incoming approach ﬂow between the time it enters the inlet and reaches the bell is a tremendous advantage over conventional wet pit pump intakes. Model baffles REFERENCES 1. and is designed in an economical manner. (e) Surface vortices were highly dependent on water levels. based on bell diameter. length. which. further veriﬁcation may be required for these cases. an internal geometry consisting of a centre ﬂoor splitter. E. are: width. and number of turning vanes equally spaced across the front entrance. it was determined that the ﬁnal FSI design is effective at straightening cross-ﬂow before it reaches the impeller. 7. and maximum pre-swirl values were well within the acceptance criteria. 2d. 2·5d. The clearance between the pump bell and the hole in the top of the inlet should be minimised with a maximum space of 1·25 cm (0·5 in) as shown in Fig. 9. Figs 5. Based on the semi-theoretical and empirical considerations for the design of this FSI. E. the ability to prefabricate the prototype FSI and simply lower it into place will greatly reduce the down-time for correcting the hydraulic problems occurring in the pump sump. (c) The fact that this inlet design is completely composed of straight pieces with the exception Formed Inlets of the curved back wall makes it very advantageous for construction purposes and therefore costs.
FLETCHER B. 1998. F. 1999. Alternate Formed Suction Inlet Design.uk. pp. 5. 1992. American National Standard for Pump Intake Design. E. NJ. 1053– 1063. Journal of the Hydraulic Division. fax or post your discussion contributions to the secretary by 1 March 2005: email: emma. E. USA.. Proceedings of the 1989 National Conference on Hydraulic Engineering. Department of Civil Engineering. New Orleans. or post to Emma Holder. P. ASCE. No. Proceedings of the ASCE International Water Resources Conference. USA. Parsippany. CHEEK D. 875– 880. 8. 6. Institution of Civil Engineers. L. Modeling in design of pumping pits. Please email.holder@ice. 1^7 Great George Street. ANSI/HI 9. Geometry Limitations for the Formed Suction Intake. fax: +44 (0)20 7665 2294. TULLIS J.8–1998. 1989. Washington. LEHR V. 9. Proceedings of the 3rd Joint ASCE/ASME Mechanics Conference Pumping Machinery. 4. University of California.. DC. Seattle. LEECH J. 2002. C. London SW1P 3AA. and HOLMAN W. US Army Corps of Engineers. Formed suction inlets on large high speciﬁc speed pumps. Hydraulics Institute. Optimizing the design of a formed suction intake for large ﬂood relief pumps. 158 Water Management 157 Issue WM3 Design guidelines for alternative formed suction inlets Werth Cheek . R. Journals Department. 7. 137– 140. HYDRAULIC INSTITUTE (HI). 1989. 3. P. ASCE. 105.org. 1110-2-327. and OSWALT R. Model study of new Madrid pumping station.2. Engineering Technical Letter No. DEMLOW T. WERTH D. USA. and CORNMAN R. Masters Thesis. Clemson University. ANTUNES F. 1979.