Experimental Testing of an Autonomous Underwater Vehicle with Tunnel Thrusters

Transcription

1 First International ymposim on Marine Proplsors smp 9, rondheim, orway, Jne 29 Experimental esting of an Atonomos Underwater Vehicle with nnel hrsters Alistair Palmer 1, Grant E. Hearn 1, Peter tevenson 2 1 chool of Engineering ciences, University of othampton, othampton, UK 2 Underwater ystems Laboratory, ational Oceanography Centre, othampton, UK ABRAC Atonomos Underwater Vehicles (AUVs) are a rapidly developing technology. his development incldes the creation of mlti-prpose vehicles capable of combining efficient long range srvey missions with low and zero speed interaction with the environment discovered. Crrent srvey-style AUVs employ control srfaces for manoevring control at speed. he ineffectiveness of these control srfaces at low and zero speeds means that additional control devices are reqired. In this paper the throgh-body tnnel thrster is considered as the additional control device. Existing pblished information abot these thrsters is reviewed and a new experimental programme is ndertaken. his experimental programme is designed to characterise the performance of both forward and aft monted tnnel thrsters in a torpedo-shaped AUV hll form at model scale. he reslts of the experimental programme are presented to show the variations in the force and moment indced on an AUV as a fnction of the operational conditions. Finally these reslts are sed to develop a simple and easily applicable modelling procedre to aid in the design of AUV control systems and for se in AUV performance analysis. Keywords Atonomos Underwater Vehicle, Modelling, nnel hrster, Experimental echniqes. 1 IRODUCIO Atonomos Underwater Vehicles (AUVs) are a class of nderwater vehicles which operate independently of any hman control. hese vehicles are controlled by onboard control systems which se the information recorded by sensors to determine the demands to be sent to the actators. he complexity of these control systems is a fnction of the sensors and actators employed and the desired vehicle performance. Another featre of the independence of AUVs is the limitation imposed by the finite energy spply which can be carried onboard. his means that design choices are heavily inflenced by the associated energy cost. here are many different types of AUV crrently in se and most of these vehicles are designed for a specific prpose. Example applications of AUVs inclde oceanographic srveying, mine-sweeping and pipeline inspection. hese widely differing applications give rise to a large nmber of different vehicle shapes and sizes, sensor packages and actator configrations. As the performance of these vehicles improves, so the desire increases to se these vehicles in a larger nmber of scenarios. herefore, the next stage in the development of AUVs is the creation of a mlti-prpose vehicle capable of combining long range srvey missions with low speed interaction and investigation style tasks. 1.1 rvey-tyle AUVs One of the first types of AUV developed was the srvey vehicle. rvey vehicles ndertake long range missions to investigate an area of the ocean and tend to carry sensor packages to facilitate scientific research. A key performance indicator for a srvey vehicle is the range it can achieve. his parameter allows an assessment of the design of the vehicle incorporating the proplsive efficiency and the energy storage capacity. herefore the design of srvey vehicles focses on combining a hydrodynamically shaped hll form and a high efficiency proplsion system with the ability to carry sfficient energy alongside the mission dependent payload. his design process has reslted in a common srvey vehicle design comprising a torpedo-shaped (or similar) hll form with a stern monted propeller and control srfaces to provide control at speed. rvey vehicles tend to be ballasted to be positively boyant to ensre that the vehicle rises to the relative safety of the srface shold the proplsion systems fail. o overcome the positive boyancy at srvey speeds, the vehicle operates at a small (nose-down) pitch angle, controlled by the hydroplanes, to generate a downwards force hydrodynamically. As a srvey vehicle slows down a speed limit is reached beyond which the control srfaces can no longer provide sfficient force to maintain the pitch angle reqired to control the positive boyancy. At even lower speeds the

2 control srfaces become ineffective and cannot provide sfficient forces to manoevre the vehicle in the desired manner. hs the creation of a mlti-prpose vehicle reqires additional control devices to provide low speed boyancy and manoevring control. he majority of nderwater vehicles se propeller based thrsters to provide low speed control de to their reliability, responsiveness and ability to generate forces throghot the speed range. o attempt to retain the srvey efficiency of the vehicle these thrsters can be monted in throgh-body tnnels. An example thrster configration for a mlti-prpose vehicle is shown in Figre 1. Vertical nnel hrsters Horizontal nnel hrsters Figre 1 Example mlti-prpose AUV based on a srveystyle configration with for additional tnnel thrsters he vehicle shown in Figre 1 has a stern monted propeller with control srfaces attached to the dct in the propeller race. In addition the vehicle has for tnnel thrsters, two in the vertical plane and two in the horizontal plane. hese additional thrsters provide control over sway, heave, pitch and yaw motions. he vehicle shown in Figre 1 will need to se the tnnel thrsters for two key tasks. hese tasks are low and zero speed manoevring control and the control of the positive boyancy at speeds below the limit of control srface control. his provides two different operational envelopes for the tnnel thrsters, namely, low and zero speed operation at a wide range of vehicle orientations and at higher speeds with a limited range of pitch angle. he exact limits of these ranges are a fnction of the design of the particlar vehicle and control srfaces. In this paper the focs is on the latter operational envelope, that is, the performance of a tnnel thrster on a vehicle moving with a forward speed. For simplicity the analysis is restricted to zero angle of attack (pitch or yaw angle, by symmetry). Initially, it is assmed that the performance of the thrster will be consistent over the range of small angles of attack experienced at speed. his is a sbject for ftre investigation. 2 UEL HRUER PERFORMACE imple design calclations for an AUV tnnel thrster dictate that the diameter, D, of the reqired thrster is small in relation to the length of the tnnel, L. he length to diameter ratio for the tnnel thrster is therefore large, for example, L/D ~ 4. he scale of these thrsters and their large length to diameter ratio means that they are nlike most marine thrsters and ths experiments have been ndertaken to characterise their performance. 2.1 nnel hrster Performance at Zero peed he performance of small diameter tnnel thrsters has been investigated nder static conditions in McLean (1991) and Cody (1992) (D = 7.5cm; L/D = 3.5 and 5.5). hese experimental reslts were sed to develop a model of the dynamic performance of the thrster (Healey et al 1995). Here dynamic refers to the performance response of the thrster sbject to a varying inpt signal. hese experiments demonstrate the steady state performance of these devices to be similar to other propeller based thrsters, that is, the thrst generated is proportional to the sqare of the rotational speed. he inflence of the long tnnel is seen in the dynamic performance of the thrster which is affected by the relatively large body of water within the tnnel. 2.2 nnel hrster Performance on a Moving Vehicle he performance of tnnel thrsters in a fll scale AUV was investigated experimentally in anders & ahon (22). he test matrix inclded the fll range of operational vehicle speeds (p to 4m.s -1 ) and yaw angles p to ±9º. However, for these experiments the thrster was isolated from the hll form and the forces recorded were those generated by the thrster and not those experienced by the vehicle. he reslts showed a variation in thrst of arond 15% over the range of forward speeds tested at zero yaw. A frther set of experimental reslts giving the performance of a tnnel thrster on a sbmersible, in terms of the forces experienced by the vehicle, is shown in Figre 2 (Beveridge 1972). K F, K Figre 2 Force and moment data for a tnnel thrster on a sbmersible (Beveridge 1972) Figre 2 ses the following coefficients to represent the performance of the thrster, see Eqation 1. A force coefficient, K F, gives the ratio of the force experienced by the vehicle to the corresponding zero speed thrster force. A moment coefficient, K, gives the ratio of the moment experienced by the vehicle to the corresponding zero speed moment. hese coefficients are plotted against the speed ratio of the vehicle forward speed,, to the thrster jet exit speed, j. F ( ), n K = (1a) F F, n ( ) KF K

3 K j (, n) (, n) = (1b) = (1c) F (, n) ρa Here F is the force on the vehicle cased by the thrster, is the moment on the vehicle indced by the thrster, n is the rotational speed of the thrster, ρ is the density of water and A is the cross sectional area of the thrster. he data shown in Figre 2 shows a variation in force of p to 93%, which is considerably more than that recorded in anders & ahon (22). his indicates that the variation in force is not solely de to a change in the performance of the thrster nit itself. 2.3 Lateral hrsters on rface Vessels he performance of a tnnel thrster on an nderwater vehicle is analogos to the performance of a lateral thrster on a srface vessel. he performance of lateral thrsters has been investigated by, amongst others, ienhis (1992), English (1972), Brix & Bssemaker (1973) and Chislett & Björheden (1966). his research incldes measrements of the forces and moments on varios vessel types, simple flow visalisation experiments and pressre measrements arond the thrster tnnel. he experimental reslts for a lateral thrster on a model tanker from Chislett & Björheden (1966) are shown in Figre 3. hese reslts show similar trends to the sbmersible data in Figre 2, inclding a large decrease in effective force with increasing speed ratio. However, there are differences inclding a sbstantial recovery of the effective moment at higher speed ratios and a higher minimm effective force. K F, K KF K Figre 3 Force and moment data for a lateral thrster on a tanker (Chislett & Björheden 1966) he srface vessel reslts, and the conclsions drawn, provide an insight into the reasons behind the variation in the performance of a lateral thrster. he thrster itself can be considered as a jet prodcing device. Hence when the vessel is stationary the jet flows away from the vehicle, see Figre 4 (dashed lines represent the path of the jet). However when the vehicle is moving forwards the thrster jet is deflected as a fnction of the relative strength of the jet to the strength of the ambient flow. As the thrster jet flow develops, flid is entrained into the jet, casing a sction effect arond the jet. When the jet is deflected backwards this sction region interacts with the vehicle, indcing a force on the vehicle opposite to the desired thrster force, see Figre 5 (shaded area represents the sction effect). he offset of this sction force from the thrster force cases a frther variation in the moment experienced by the vehicle. (ew symbols sed on Figres 4 and 5 are defined later, where appropriate). z x x Figre 4 chematic of tnnel thrster performance at zero speed z x x x F F Figre 5 chematic of tnnel thrster performance on a vehicle moving with a forward speed his representation of the operation of a tnnel thrster ignores the inflence of sction effects pstream of the tnnel otlet and arond the tnnel inlet. Pressre measrements arond the inlet and otlet of the tnnel have shown that the primary case of the loss of effective force is the inflence of the sction downstream of the tnnel otlet (English 1972; Brix & Bssemaker 1973). he complexity of the interaction between the ambient flow (inclding the bondary layer), thrster jet and vehicle mean that the performance of each different thrster configration is niqe. herefore to be able to characterise the performance of a tnnel thrster on an AUV type body an experimental approach was adopted. 2.4 AUV imlation imlations are commonly sed in the development of AUVs to aid in control system design and to gain insight F Flow

4 into the performance of the vehicle. For the simlations to accrately reflect the performance of the vehicle it is necessary to model the inflence of the actators employed. However, no common modelling approach for the performance of a tnnel thrster is readily available. his is thoght to be de to the complexity of the interactions involved and the niqeness of each configration. Pblished AUV simlations tend to assme that the forces experienced by the vehicle are eqal to those generated by the thrster and that the indced moment can be calclated according to geometric considerations (for example, Ananthakrishnan et al (1998)). anders & ahon (22) did modify the existing Healey et al (1995) thrster model sing a look-p table of experimental reslts. However, as the experimental reslts do not accont for the ambient flow effects, the model does not represent the complete performance. Hence the reslts obtained from the experiments ndertaken will be sed to develop a modelling approach for the performance of a tnnel thrster on an AUV. 3 EXPERIMEAL EUP A 2.5m, approximately one-third scale, model of the srvey AUV Atosb 3 (Fallows 24) was modified to accommodate two horizontal throgh-body tnnels, one forward and one aft, as shown in Figre 6. he tnnel locations were selected by considering the vehicle performance across the entire operating envelope and the existing se of internal space. Each tnnel has a diameter eqal to that of the thrster monted within the tnnel and the tnnels are symmetrically faired into the shape of the vehicle at the inlet and otlet. he particlar thrsters sed are 7mm diameter rim driven thrster nits (Ab- harkh et al 23). hese thrster nits are well sited to this application as they provide symmetrical performance and minimise the blockage in the tnnel cased by the hb. he thrsters were driven sing an electronic speed controller connected to a compter. he rotational speed of the thrster was measred by monitoring the control signal sing a freqency conter and the electrical power drawn was recorded. monting poles, onto a prpose designed and bilt dynamometer which incorporates for force blocks (Fallows 24). Each force block ses a linear variable differential transformer to measre the transverse displacement indced by a force applied between the top and bottom of the block. he force blocks are monted in orthogonal pairs on each monting post to measre drag and side force. Each force block was calibrated sing a mlti-point calibration sing calibrated weights. he force block signals were digitised and passed to a compter for atomatic data logging. he data was recorded at 6Hz. he test programme for the experiments was designed to cover the range of operational conditions expected and was ndertaken in stages. Firstly, the drag of the vehicle was measred over the fll range of carriage speeds (p to 4.25m.s -1 ). Following this the thrster performance was characterised at zero speed over the range of rotational speeds to be sed in the testing. Finally the vehicle was rn with the thrsters operating at selected rotational speeds (to give an appropriate range of speed ratio) sing for different vehicle speeds approximately.4m.s -1,.6m.s -1,.8m.s -1 and 1.m.s -1. he reslts of these tests are detailed in the following sections. 4 REUL he drag of the vehicle was recorded, withot the thrsters operating, to assess the srvey drag impact of adding thrster tnnels to an AUV hll form. he increase in the drag of the vehicle, with thrster tnnels compared to withot thrster tnnels, at srvey speeds, was less than 2%. 4.1 hrster Performance at Zero peed he force generated by the thrster, F, at zero vehicle speed was recorded throghot the range of rotational speeds, n. hese reslts are illstrated in Figre 7 showing a linear trend of thrster force with the sqare of the rotational speed as expected. Frthermore, these reslts closely match the data pblished by the manfactrer for the thrster operating with its standard dct fitted (L 26). 4 3 F () 2 Figre 6 CAD drawing of the Atosb model showing forward and aft thrster tnnels (control srfaces and stern proplsor not shown) he tank sed for the testing was the othampton olent University owing ank which is 6m long, 3.7m wide and 1.85m deep and has a carriage which can rn p to 4.25m.s -1. he model was monted, via a pair of n 2 (s -2 ) Figre 7 nnel thrster performance at zero speed showing thrst, F, against the sqare of the thrster rotational speed, n 2

5 K F, K Aft KF Fwd KF Aft K Fwd K Figre 8 nnel thrster performance for forward (hollow symbols) and aft (solid symbols) thrsters on an AUV moving with a forward speed 4.2 hrster Performance on a Moving Vehicle he reslts for the forward and aft thrsters across the range of speed ratio are shown in Figre 8. he data recorded is presented sing the coefficients defined in Eqation 1. he forces indced by the operation of the thrster are determined by calclating the difference between the forces recorded at a given speed with and withot the thrster operating. hese reslts show a large drop off in the force experienced by the vehicle with increasing speed ratio for both the forward and aft thrsters. his data shows a similar form to that presented in Beveridge (1972) (Figre 2). he differences between the force trends for the forward and aft thrsters are an indication of the inflence of the differing form of the vehicle arond the two tnnel exits. he hll form aft of the rear tnnel slopes away from the tnnel whereas the hll form aft of the forward tnnel is flat. his variation will give a differing interaction between the deflected jet and the vehicle and as a conseqence reslts in differing performance characteristics. Figre 8 also shows that the moments indced by the thrsters drop off considerably with increasing speed ratio. However, there is a notable difference between the variations for the two thrsters. In order to gain some insight into these variations it is necessary to nderstand the forces acting on the vehicle. A simplified representation of a tnnel thrster ses two forces (see Figre 5). hese are the thrst force generated, assmed to act at the thrster axis, and a total sction force, which acts at a variable point as a fnction of speed ratio. he sction force, F, is defined as the difference between the expected (zero speed) force and the force experienced by the vehicle, see Eqation 2. he sction moment,, is defined as the difference between the expected (zero speed) moment and the moment experienced by the vehicle. F, n = F, n F n (2a) ( ) ( ) ( ) (, n) (, n) ( n), = (2b), he centre of action of the sction force, x, is then defined as the ratio of the sction moment to the sction force: ( ), n x (, n) =. (3) F, n ( ) Figres 9 and 1 show the variation of the centre of action of the sction force, with speed ratio, for the forward and aft thrsters, respectively. x (m) Figre 9 Centre of action of the sction force, x, for the forward tnnel thrster against speed ratio, hese reslts show that the centre of action of the sction force for the forward thrster moves aft (towards zero) with increasing speed ratio. his movement is towards the central pivot point of the vehicle and ths redces the

6 impact of the sction moment, giving the limited redction in moment shown on Figre 8. For the aft thrster the centre of action is roghly constant and aft of the thrster, giving a relatively greater inflence of the sction moment. his leads to the point at a speed ratio of approximately.5, where the aft thrster moment coefficient changes sign as the sction moment dominates the moment created by the thrster force. he reason behind this relatively constant centre of action for the aft thrster is thoght to be de to the trncation of the hll form downstream of the tnnel exit. x (m) Figre 1 Centre of action of the sction force, x, for the aft tnnel thrster against speed ratio, Both thrsters were rn simltaneosly, in the same and opposite directions, and no interaction effects were experienced at the large thrster separation sed. hs, the performance of the two thrsters combined is eqal to the sm of the performance of the individal thrsters. 4.3 hrster Indced Drag he drag force on the vehicle was recorded dring the thrster experiments. Figre 11 shows the increase in volmetric drag coefficient, C D, compared to the thrster-off case, against speed ratio. C D Aft Forward Figre 11 Increase in volmetric drag coefficient, C D, against speed ratio,, for the forward (hollow symbols) and aft (solid symbols) thrster operation hese reslts show that the increase in drag redces as speed ratio increases and the variation is the same for both thrsters. At the low speed ratios, where the jet dominates the ambient flow and effectively forms a cylinder in the flow, the increase in drag is the largest. As the speed ratio increases, and the jet is deflected more by the ambient flow, the increase in drag redces. ote shold be made that the vehicle drag at low speeds is small and hence a large increase in drag coefficient does not correspond to a prohibitive increase in actal force. 5 UEL HRUER MODELLIG Accrate simlation of the performance of an AUV eqipped with tnnel thrsters reqires a model of how the operation of these thrsters affects the vehicle. ince no established modelling procedre is readily available, the data obtained from these experiments will be sed to develop a simple and easily applicable model. Yoerger et al (199) state that at low speeds the control of an AUV can be dominated by the dynamics of the thrsters employed. herefore it is important to inclde the dynamic effects of the thrster in the modelling procedre. anders & ahon (22) conclded that the dynamic performance of the tnnel thrster tested was nchanged by the range of experimental conditions experienced. his conclsion is backed p by a series of dynamic experiments ndertaken sing the experimental setp tested here. herefore, existing models of the dynamic performance can be employed, for example, Healey et al (1995). o model the steady state performance of the thrster an exponential can be fitted to the force reslts of the form: 2 K F = exp c. (4) j he force experienced by the vehicle can readily be determined by applying a model of the performance of the thrster at zero speed. he selection of the constant, c, is a fnction of the individal configration tested. Using a least sqares approach, c 7 for the forward thrster and c 3 for the aft thrster. o model the moment experienced by the vehicle, the simplified representation consisting of only two forces is sed. hs, the moment is given by: = F x + F x. (5) he thrster force, F, is determined from a zero speed thrster performance model and the thrster moment arm, x, is determined sing the geometry of the vehicle. he sction force, F, is determined from Eqation 2 and the sction moment arm, x, is represented sing a simple model. Chislett & Björheden (1966) conclde that the centre of action of the sction force moves linearly aft with increasing speed ratio. he reslts for the forward thrster, presented in Figre 9, show some agreement with this conclsion, giving a model of the form: x = x kd. (6) Here k is a configration dependent constant. However the reslts for the aft thrster do not follow this linear trend and hence a constant vale of x is applied here. hese reslts show that it is important to accont for the j

7 shape (and trncation) of the body when selecting the model strctre sed for the sction moment arm. 6 COCLUIO o create a mlti-prpose AUV capable of ndertaking both srvey-style missions and low speed interaction with the environment encontered reqires the addition of frther control devices to agment common srvey AUV configrations. o retain the existing srvey efficiency one available option for these additional control devices is the throgh-body tnnel thrster. his paper reviews the available pblished data for the performance of tnnel thrsters on AUV type bodies and finds a need for additional experimental testing. herefore experiments were ndertaken sing rim driven thrsters monted in fore and aft throgh-body tnnels on a model scale torpedo-shaped AUV. he reslts of these experiments are presented to show how the forces and moments experienced by the vehicle, de to the operation of the tnnel thrster, vary as a fnction of the operational conditions. hese reslts show considerable decreases in the force and moment indced on the vehicle by the thrster with increasing speed ratio. hese reslts also show how the variations in the shape of the vehicle arond the tnnels can lead to sbstantially differing performance. o aid in control system design and AUV performance analysis, a simple and easily applicable modelling procedre for the performance of a tnnel thrster on an AUV type body, as determined from the experiments, has been presented. he work reported here forms part of a wider research project investigating approaches to extending the capabilities of srvey-style atonomos nderwater vehicles. ACKOWLEDGEME he athors wold like to thank L echnology Ltd for the loan of the rim driven thrster nits sed in the experiments, the technical staff in the chool of Engineering ciences at the University of othampton for their help in setting p the experiments and the Engineering and Physical ciences Research Concil (UK) for financial spport. REFERECE Ab-harkh,.M., rnock,.r. & Hghes, A.W. (23). Design and Performance of an Electric ip- Driven hrster. Jornal of Engineering for the Maritime Environment, Proceedings of the Instittion of Mechanical Engineers, 217(M), pp Ananthakrishnan, P., Vantorox,., Zhang, K. & mith,. (1998). onlinear Dynamics and Hydrodynamics Considerations in the Design of an Atonomos Underwater Vehicle. Proceedings of Oceanology International 1998, Brighton, UK, pp Beveridge, J.L. (1972). Design and Performance of Bow hrsters. Marine echnology, 9(4), pp Brix, J.E. & Bssemaker, O. (1973). Lateral hrsters with Anti-ction nnels. Proceedings of the 1 st orth American g Convention, hip & Boat International, Vancover, Canada, pp Chislett, M.. & Björheden, O. (1966). Inflence of hip peed on the Effectiveness of a Lateral-hrst Unit. Report Hy-8, Hydro-og Aerodynamisk Laboratorim, Lyngby, Denmark. Cody,.E. (1992). Experimental tdy of the Response of mall nnel hrsters to qare and rianglar Wave Inpts. Masters hesis, aval Postgradate chool, Monterrey, CA, UA. English, J.W. (1963). he Design and Performance of Lateral hrst Units for hips. ransactions of RIA, 15(3), pp Fallows, C.D. (24). Characterisation of the Proplsion ystems of Underwater Vehicles. PhD hesis, University of othampton, othampton, UK. Healey, A.J., Rock,.M., Cody,.E., Miles, D. & Brown, J.P. (1995). oward an Improved Understanding of hrster Dynamics for Underwater Vehicles. IEEE Jornal of Oceanic Engineering, 2(4), pp McLean, M.B. (1991). Dynamic Performance of mall Diameter nnel hrsters. Masters hesis, aval Postgradate chool, Monterrey, CA, UA. ienhis, U. (1992). Analysis of hrster Effectivity for Dynamic Positioning and Low peed Manoevring. PhD hesis, Delft University, Delft, etherlands. anders, A. & ahon, M. (22). he Effect of Forward Vehicle Velocity on hrogh-body AUV nnel hrster Performance. Proceedings of Oceans 22 Conference, M/IEEE, Biloxi, MI, UA. L echnology Ltd. (26). Integrated hrster Data. (Viewed 3 Jan 29). Yoerger, D.R., Cooke, J.G. & lotine, J.-J.E. (199). he Inflence of hrster Dynamics on Underwater Vehicle Behavior and heir Incorporation into Control ystem Design. IEEE Jornal of Oceanic Engineering, 15(3), pp