To demonstrate this last capability, a low budget simulator has been built for demonstration purposes. This article describes the simulator, the development of the mathematical models and the implementation in a micro computer.
Introduction
In the second part of the 1970's, the first remotely operated underwater vehicle (ROV) was introduced in the North Sea. Since that time, the number of vehicles has shaiply increased, as is the number of available types. I n fact, there are only a few types which have seen their production number increase above 10. This situation is now slowly evolving; the days of the one-of-a-kind all purpose ROV are num-bered. The new ROV's are designed with a specific task in mind. These tasks are also getting more dif-ficult in the near future, and the requirements on ROV task performance are increasing.
A t TNO-IWECO, we are convinced that math-ematical simulation can be a valuable tool to f u l f i l these more stringent requirements. Simulation of the manoeuvring behaviour can be used in the design stage, to assess the task capability of the proposed design. I f the task is very difficult and i f there is much money at stake, a special training simulator can be
*) TNO-I\VECO, Delft, Holland..
used to good effect.
As a first step towards the development of these simulation tools, we have developed a low budget simulator for demonstration purposes. We have select-ed a Dutch ROV to simulate its manoeuvring charac-teristics. The simulator consists of an operator con-sole, a computer and the software. The software in-cluding the mathematical models is implemented in a micro computer. The micro computer is installed in the back of the console; in this way the simulator is very compact and transportable. The simulator was
developed by a four man team in three months; i t was first shown on a symposium to celebrate the 30th anniversary of TNO-IWECO in May 1984.
This publication describes the simulator.
The simulated ROV
The simulated ROV is a small vehicle, built and operated by Heerema, the SUB 300, Figure 1. The SUB 300 is mostly used for pipeline inspections and comparable tasks; itg main characteristic is its
suit-Manufacturer Heerema Innovation Eng., Leiden, The Nether-lands
BuUd year 1984
Dimensions 0 1.25 m x 1.00 m Weight, dry 500 kgf
Pay load 80kgf
Structure open frame with cast aluminium doughnut-shaped pressure housing on top
Thrusters 4 rpm-controlled thrusters in hor. X-Y con-figuration, 1 vert, thruster in doughnut center Power thruster power 5 x 5.5 kW.
g a r a g e
Figure 2. ROV system witli surface support vessel, umbilical, garage, tether and ROV.
f l o a t s
w e i g h t s
Figure 3. ROV system with surface support vessel, umbilical, ROV.
ability for operations at high current velocities. The thmster-power/weight ratio is very favourable, and, also very important, the thrust is high in every horizontal direction.
Another basic feature of this ROV is that i t is operated from a garage. Figure 2. Using this system, the connection to the surface is not limiting the free-dom of the ROV in a way as with the older system of one continuous umbihcal to the surface support ves-sel. Figure 3.
EL
p o s i t i o n v e l o c i t y ( k ) p o s i t i o n f o r c e s on ROV ( g r a v i t y ) 0-v e l o c i t y f o r c e s on ROV ( h y d r o d y n ) a c c e l e r a t i o n f o r c e s on ROV, "added mass'5
f o r c e s o n u m b i l i c a l
EL
t h r u s t e r f o r c e s
The ROV is equipped with video-cameras (colour, black and white), sonar, lights and an automatic head-ing and depth control system.
The mathematical models
The flow diagram of the mathematical models in the simulation program is given in Figure 4. As this diagram shows, there are several modules which calculate external forces. These forces are substituted in the 6 degrees of freedom equations of motion. An integration routine completes the simulation loop.
7 6F e q u a t i o n s o f m o t i o n
0-i n t e g r a t 0-i o n o u t p u t'black-out'. This positive buoyancy force {FG-FB) is transformed to the ROV system of axes by multi-plication with the rotation matrix R. The rotation matrix R is defined to transform vectors from the earth-fixed system of axesZ^F^Z^ to the body fixed
system XYZ. j module 3
The velocity forces are constructed from h f t and drag data of basic elements. These h f t and drag forces are based on data in Hoerners work [ 5 , 6 ] . The forces on the basic elements of the ROV are added; inter-action forces and shadowing effects are estimated.
In this way, the hydrodynamic forces are calculated for a limited number of cases (a, (3); intermediate values are interpolated.
Diagrams showing the forces as a function of OL are given in Figure 5.
In formula:
+ Vzp CO |aj|
in which:
CO - angular velocity of vehicle.
Coefficients XAC^{a,P) and S ^ q , ( a , ( 3 ) contain the sum of the products of relevant area and drag/lift coefficient of each basic element including interac-tions. The coefficients can be found in look-up tables.
module 4
Apart from the virtual increase of the mass of a sub-merged vehicle, the 'added mass' phenomenon has more effects. These effects can be determined i f the equations of motion are derived for a moving system of axes with the 'total mass' matrix (total mass = mass + added mass). This derivation was, f o r instance, pub-lished by F.H. Imlay [ 7 ] . For the X-equation, this results in:
F^acc = « + {w + uq) + X.q+ Z . wq+Z.q^ + + X.v + X.p + X.f - Y,vr - Y.rp - Y.r^ + - X ur - Y. wr + Y^.vq + Z.pq - (Y. - Z.) qr The symmetry of the added mass matrix has been used in this expression. Any symmetry of the
con--0.5 •1 .0 - 1 . 5 -2.0 -2.5 a
\
/
M(a) Z(a) - 0 . 5 •1.0Figure 5. Velocity (lift and drag) forces on the SUB 300 as a function of the vertical angle of incidence a. Y(a) = K{a) =
N(a) = 0.
sidered vehicle is not included. Even i f the vehicle has only one plane of symmetry, this expression is sig-nificantly reduced. For this ROV, which is nearly symmetric with respect to the Z-axis, we used:
module 5
For the calculation of the forces acting on the bilical, and most important f o r the forces of the um-bilical on the ROV, we have developed two models.
Model 1 uses a 'finite segment' approach: the um-bilical is divided in a number of segments. Each of these segments may be curved, but has its mass con-centrated in the center. This point is also the center of all extemal forces. This model gives a fairly ac-curate description of the whereabouts of the umbihc-al and of the forces exerted by the umbihcumbihc-al on the ROV. I t is, however, not very suitable for real-time simulation on the applied micro computer.
For our simulator, we needed a more simpUfied model of the umbihcal. Since our ROV uses a cage-tether system. Figure 2, i t seems to be acceptable to use a quasi-static approach.
With this approach, the tether is assumed to have a parabohc shape. The parabola is defined by the locat-ions of ROV and garage and the length of the tether. The plane of the parabola is defined by the relative velocity vector of the tether through the fluid and the line which connects the ROV and the garage.
The dynamic simulation of the tether motions is thought to consist of a series of static situations.
module 6
The basis of this module is the four-quadrant look-up table of the thrusters. This table gives thrust and torque coefficients of the thrusters for values of the hydrodynamic pitch angle ^:
0 < (3 < 360 degr.
This hydrodynamic pitch angle (3 is defined: tan|3 =
OJirnD
in which:
u - inflow velocity n - revolutions per second D - diameter o f propeller. The four quadrants are defined:
0 < /3 < 90 degr. u> 0 n>0 90 < p < 180 degr. u> 0 « < 0 180 < (3 < 270 degr. u<0 n<0 270 < |3 < 360 degr. u<0 n> 0 The formulas f o r thrust and torque are: T=C,
4
Q = C* • V2p[u^+ (0.7n nD)^]^D^
The values of the coefficients C* and C* are found in the look-up table. A continuous graph of this table is given in Figure 6.
The thruster module calculates the local flow con-dirions for each thruster, the thrust and torque, and the thruster outflow velocity. I f necessary, the thrus-ter rpm is reduced i f the maximum available power is exceeded. The thruster outflow velocity is used to calculate the interactions between the four horizontal thrusters. The outflow velocity is calculated consider-ing the propeller as an actuator-disk:
^ u t = 2 s i g n ( r )
" - - + 1 /
jru- \ 1^1 1 2 1/ 2_in which:
A^f^ - thmster disk area.
This outflow velocity influences the local flow ditions of a neighbour thruster. These local flow con-ditions, inclusive interactions, determine:
- 0 . 5 u > 0 n > 0 u > 0 n < 0 u < 0 n < 0 u < 0 n > 0 0 A / \ 90 180
/\
360r
/
\/
Figure 6. Four-quadrant diagram of the propeller type of the SUB 300. AU five propellers are idential: P/D = 0.85, Ae/Ao = 0.50,£) = 0.25.
— propeller inflow total velocity
— angle of incidence of inflow total velocity.
The final thrust and torque of the propeller are then calculated. The influence of the angle of incidence of the inflow velocity on thrust, torque and sideforce production is taken into account by a look-up table based on experiments [ 1 1 ] . A continuous graph of this table is given in Figure 7.
The vertical thruster is a special case. This thruster is situated at the center of the buoyancy volume, and cannot be considered as a thruster in a nozzle at an angle of attack. The sideforce production calculation of the four horizontal thrusters is changed for a 'momentum drag' calculation.
I f the ROV is flying forward, and the vertical thruster is operated, the flow is to be forced from the horizontal plane through the vertical thruster. The reaction of this force on the ROV is called momentum drag. A formula for the momentum drag of the des-cribed case is:
F in which:
dm , ,
^pd ' velocity through propeller disk. This velocity can be estimated as Vir^^^^j.
The thrust of this vertical propeller is calculated in the way as is described for the horizontal thrusters.
module 7
This module solves the equations of motion; the output contains the accelerations at time level k.
module 8
Module 8 contains the exphcit integration formulas.
The simulator program
The simulation program as given in Figure 4 is ex-tended with several additional routines to make it suitable f o r a simulator. These additional routines are: - analog/digital input
- analog output
- thruster aUocation logic (to translate the joystick position and signals from the heading control to commanded thruster rpm's)
and for our special case - image generation - heading control.
The analog and digital conversion routines are machine-dependent and are not described in this publication.
The thruster allocation logic is straight forward and also not described; some more attention is given to the last two routines:
I?72age generating routine
Due to the hmitations of our graphics processor (it takes 800 ns to put 1 pixel on the monitor), we de-signed a simple computer generated image. The sea bottom is represented as a perfect horizontal plane with a 1 X 1 m^ grid on i t . This grid gives the pilot a feeling of horizontal speed and changes in course.
Changes in depth are indicated by an increase/ decrease in size of the grid on the monitor and a change i n the perspective of the view. The range of
visibility is adjustable, we mostly use a range of 4.0 m. Our simulator hydrospace is, apart from the bottom, limited by one vertical subsea wall. This wall enables the pilot to practice vertical fUglit,
The simulation program has to run real time for the simulator. In order to present a smooth picture on the monitor, the update frequency has to be rather high. We achieved 8 Hz on our present micro computer. This update frequency gives an almost smooth im-pression of motions on the monitor.
Heading control
The simulated ROV is rather high powered and nearly axi-symmetric with respect to the Z-axis. The combination of these qualities makes i t very difficult to keep the ROV manually on course. Therefore, a heading control system is incorporated in the simul-ator.
The control system is of the PID-type; i t makes use of a difference i n course, a course rate of turn and an integrated course difference to determine the com-manded moment. The thruster allocation logic determ-ines thruster rpm's from this commanded moment.
The control algorithm reads;
* ) + 2 K.^t
( * setp setp
This control algorithm is supported by several routines to guarantee a smooth initiahzation (from manual to automatic mode) and a controlled rate of turn i f a new setpoint is entered.
The selected setpoint can be adjusted with ± 3 0 degrees by the third axis, rotation, of the joystick.
The microcomputer system
The microcomputer used for the simulator is the versatile Geminix system, developed by a colleague TNO insitute, TNO-IBBC. The system uses the
d i g i t a l analog system s e t p o i n t heading j o y s t i c k o p e r a t i n s y s t e m t h r u s t e r a l l o c a t i e l o g i c heading c o n t r o l heading ROV + t h r u s t e r s + rpm u m b i l i c a l image g e n e r a t i o n g r a p h i c s processor
Figure 8. Flow diagram of simulator program.
Motorola MC68000 microprocessor both as master and slave processors on a standard VMEbus.
Additional hardware consists of memory, A / D and D / A converters, parallel and serial I/O, and a graphic display controUer from NEC, the mPD7220. A hard-ware floating point module may be added to increase performance.
To reflect the modularity of the simulator system setup in the hardware configuration, a separate proces-sor was used for each software module, see Figure 8. This results in rather limited dataflow between the four processes and allows easy interchangeability of software modules. By replacing only one software module, another ROV may be simulated, but also different consoles and/or video monitors may be con-nected to the simulator in the same fashion. I t is also possible to 'isolate' a module and to develop and test i t separately.
A l l software is written in the programming language 'C' and is, apart f r o m some minor I/O drivers, trans-portable to another computer system.
The console
The console of the simulator. Figure 9, is a close rephca of the one used f o r the real SUB 300. I t is equipped with:
- 3-axes self centering joystick foiX—Y motions and yaw
- thumbwheel for the Z-motion
- 3 thumbwheels for trimfunctions of horizontal plane motions
setpoint selector for automatic heading control system
2-axis joystick for camera pan/tilt
displays of • 5 thmster rpm's (analog) ' heading setpoint (digital) • actual heading (digital)
- depth (digital) " depth rate (analog) simulator mode selector (ON/OFF, I N I T I A L CON-DITION, RUN, HOLD)
monitor for the cort)puter generated imagery.
Figure 9. Operator console of the simulator (at a development stage).
of the real SUB 300 i n pitch and roll to be higher than that i n the simulation. They concluded that a simul-ator could be very useful, even i n this basic configurat-ion, as an aid for basic pilot training.
5
Future developments
We have a strong believe that simulation tech-niques wül earn their role in underwater technology. Most logicaUy, the field of application will be the training of pilots and evaluation of designs. I f the task to be performed by the ROV is very difficult and i f there is much at stake, a special training for this task is justified.
Another possible apphcation is to aid the operator with the navigation problem inside a jacket. His com-mands to the real ROV will also be fed to the math-ematical models i n the computer. He can then use the computed view in comparison with the actual view. The computed view wUl give the advantage of infinite visibility and the pilot can zoom in or out to assess his situation.
I t should also be possible to select a mode i n which the view on the screen is the one of a spectator stand-ing outside the jacket. The monitor then gives a
pic-don, Paper 9.4.
3. Evensen, G., Det norske Ventas, 'ROV training is imperat-ive', ROV '83, San Diego.
4. Partridge, D.W., Department of Energy, U.K., 'Future developments of ROV's', Underwater Technology, Vol. 10, no. 1,1984.
5. Hoemer, S.F., 'Fluid dynamic drag', PubUshed by the author, 1965.
6. Hoerner, S.F., 'Fluid dynamic h f t ' , Pubhshed by L.A. Hoerner, 1975.
7. Imlay, F.H., David Taylor Model Basm, 'The complete expressions for 'added mass", DTMB report 1528, July 1961.
8. Lewis, D.J., Lipscombe, J.M. and Thomasson, P.G., Cran-field Institute of Technology, 'The simulation of remotely operated underwater vehicles', ROV '84, San Diego. 9. Miller, M.J., Santa Fe Underwater Services Inc., ROV
test-ing and evaluation facilities, ROV '84, San Diego.
10. Tohnan, F., TNO-IBBC, 'The Geminix workstation', CAPE '83, Amsterdam, page 923-928.
11. Beek, J.v.d., and Amersfoort, H.C. van, Netherlands Ship Model Basin, 'Experimental investigation into the hydro-dynamic characteristics of a 1975 HP thmster unit', NSMB report 0968-1-DT, November 1975.
12. Lammeren, WR.A. van. Manen, J.D. van, and Oosterveld, M.W.C., Netheriands Ship Model Basin, 'The Wageningen B-screw series', Schip en Werf, Vol. 37, no. 5,1970. 13. Thomasson, P.O., Cranfield Institute of Technology,
'Simulators for use as design aids and for operator train-ing', SUBTECH '83, London, Paper 9.1.
