INS for DP – Methods – Full Version IMCA Annual Seminar Singapore, November 2013, Jim Titcomb, Yann Casamajou
Aug 18, 2015
INS for DP – Methods – Full VersionIMCA Annual Seminar Singapore, November 2013, Jim Titcomb, Yann Casamajou
INS Introduction
3What's in an INS?
What is an Inertial Navigation System (INS)?An instrument (electronic + sensors) which is using its initial state (position) and internal motion sensors (gyroscopes + accelerometers) to measure and calculate its subsequent positions in space with high accuracy, stability and update rate
PHINS PHINS 6000 ROVINS
4What's in an INS?
3 “FOG” gyrometersmonitor rotation and speed in X, Y & Z axis
3 accelerometersmeasure acceleration (>> speed >> motion) in 3 axis
Powerful electronic / firmware packagePHINS “knows” in real time its motion in space . Firmware (Kalman filter) calculates its position in real time + heading, pitch, roll, heave, etc…
All integratedsmall, lean, powerful!(PHINS 6000 example)
5INS Introduction
What makes an INS good.. or not? Internal sensors (gyroscopes & accelerometers) are never perfect,
bias and scale factors accumulate over time Navigation is mostly about Gyroscopes
Gross figures: Accelerometers errors are not heavily involved in the position drift –
4m – Schuller period Gyroscope are heavily involved in the drift – 400 m
Conclusions A good INS requires good gyro’s IXBLUE manufactures FOG (Fiber Optic Gyroscope) and controls the
whole process A range of FOG’s (FOG90, FOG120, FOG180…) for a range of INS
6INS Introduction
The best sensors are still not perfect, accumulating small errors vs. time makes the system drift on the long term
External sensors (aiding) are required to bound drift within acceptable limits.PHINS & ROVINS includes interfaces for most common external sensors
GPS DVL (Doppler Velocity Log) Pressure sensor Acoustic positioning references (USBL, LBL) …and all IXBLUE products!
IXBLUE INS are fully integrated Inertial Positioning solutions designed for ease of installation & operation, flexible enough to fit most requirements, with no specialist engineer to install / operate.
7Benefits of data fusion: Robustness to signal losses
GPS positioning with masking
8Benefits of data fusion: outages
GPS + INS positioning with masking
9Benefits of data fusion: accuracy
Data Fusion Use various and different technologies to measure the same parameter Blend (fuse all this data (Kalman filter) in order to correlate it and obtain a
better result A simple example GPS + INS (PHINS or GAPS typical use case)
0
0,5
1
1,5
2
2,5
3
0 20 40 60 80 100
time t (s)
po
sit
ion
ac
cu
rac
y (
m)
PHINS pure inertial drif t (0.0002 x t^2m)
Averaging of DGPS data (3 / sqrt(t)m)
PHINS+DGPS
10Benefits of data fusion: accuracy (USBL case)
Acoustic positioning can be poor, low update rate, or out of range. INS + USBL combination / data fusion provides continuous high quality positioning:
Survey @2500m depth: INS accuracy is much better than USBL’s one (noise
rejection)
DP-PHINS : Bringing INS benefits to DP
12
DP-PHINS What is it?
Missing link between PHINS and USBL PME
Goals of the system :
DP-PHINS cabinet extends PHINS capabilities to :
Context: Offshore works and installation jobsDeep sea (>1000m)Necessity to pursue operations without any reliable GPSSevere daily scintillation phenomena (Africa, Brazil...)
Use PHINS fusion algorithms to enhance raw USBL positioning
Improve USBL performances to make it usable for DP during GPS scintillation
use raw acoustic detections from any USBL for positioningprovide positioning telegram to DP Desk
13DP-PHINS What is it?
Time Signals
Time and Position For Initialisation Only
Time stamped beacon positions and heading
Processed Positions
Enhanced Position
Acoustic Positions
14DP- PHINS, Features
PHINS Natural Features: Native fusion of a wide range of sensors (GPS, DVL...) Strong noise rejection Sparse-array capability
DP-PHINS cabinet additional improvements: Relative to Global co-ordinate transforms Unlimited number of beacon Sparse array LBL Flexible Expandable
Future Sensors Taught Wire Fan Beam Radascan Etc. Etc.
15Why this approach?
Augmented GNSS is an accurate and generally reliable positioning solution.
For DP in deep water, the primary issue would seem to be lack of other available positioning systems.
What to do in water that is too deep for useful acoustics? After a certain depth acoustics become too noisy, too deep for taut
wire, no other structures for relative based systems.
Too many vessels relying on single PME – GNSS
What does INS give you when you are relying on GNSS?
Data Example – Single PME
17Station keeping on GPS
181 minute outage
192 minute outage
203 minute outage
214 minute outage
225 minute outage
236 minute outage
247 minute outage
258 minute outage
269 minute outage
2710 minute outage
11m in 10 minutes
2820 minute outage
38m in 20 minutes
29Whats the alternative?
INS will drift quickly if the aiding sensor fails. 3m in 2 min, 20m in 5 min. 0.6Nmi in an hour
When using INS with a single aiding sensor, a single failure will take out your INS also.
INS is most effective with multiple aiding sensors.
INS SHOULD be used with multiple aiding sensors.
DP-PHINS : Multiple aiding sensors
31DP-PHINS: what if the positioning source fails ?
INS free inertial specification is based on time. The more time, the greater the drift.
Adding a DVL to DP-PHINS is an option to contain this drift DVL aided INS specification is based on distance travelled. If you don’t move
the error can’t grow as much.
PHINS-DVL performance is still good at low update rate compatible with deep water
Problem: DVL is only available in water depths up to around 1,000m
DVL Update Rate
Error, % travelled distance
Drift speed m/h
1S 0.03% 0.18
2S 0.13% 0.82
3S 0.26% 1.67
4S 0.27% 1.77
6S 0.32% 2.08
8s 0.30% 1.98
32DP- PHINS Performances: Augmented USBL, case 1
USBL raw dataPHINS + USBL, SD livePHINS + DGPS
Nov2012: 1300m depth operations
Single vessel, 2 cases:
DP with manoeuvring operations1x standard omnidirectional transducer, tonal codesDP with static vessel position1x directional transducer improved USBL accuracy
Accuracy Gain
x3.3
x2.65
Accuracy (%slant range)
0.055%
0.035%
Test CASE USBL SD (1DRMS,DGPS ref)
DP-PHINS SD (1DRMS,DGPS ref)
Dynamic, omni 2.34m 0.71m
Static, directional 1.22m 0.46m
DP-PHINS raises USBL PME to submetric
performances on performing vessels
33DP- PHINS Performances: Augmented USBL, case 2
Less performing USBL
Kongsberg CAT, 1360m depthStandard tonal transpondersDP performances in dynamic test
Error Type USBL vs GPS accuracy
PHINS vs GPS accuracy
max 86,79m 7,94m
1DRMS 6,21m 2,64m
Accuracy GAIN (GPS ref)
x10,93
x2,35
% Slant range
0.57%
0.19%
DP-PHINS raises poor DP-USBL to acceptable levels of performances in most fields
34Acoustic Aiding Examples - Recap
ROV on Seabed. 850m water depth Well calibrated USBL About as good as it gets
with USBL
Standard Deviation
X Y Position
USBL 1.03
0.86
1.34
35INS Aided with USBL - Recap
Spikes in USBL rejected Automatically.
Much higher update rate. Smoother positioning. Reduced spread of
positions.
5 X improvement with INS
Standard Deviation
X Y
USBL 1.03
0.86
1.34
INS 0.16
0.2 0.26
36INS Aided with USBL & DVL
DVL update only every 3 seconds
Positioning now better than high accuracy GPS even in 850m
17 x Better than raw acoustics.
Standard Deviation
X Y
USBL 1.03
0.86
1.34
INS 0.16
0.2 0.26
With DVL 0.05
0.06
0.07
37What if we lose acoustics?
DVL update only every 3 seconds
Positioning now better than high accuracy GPS even in 850m
Standard Deviation
X Y
USBL 1.03
0.86
1.34
INS 0.16
0.2 0.26
With DVL 0.05
0.06
0.07
INS DVL 0.04
0.02
0.05
DP-PHINS : Future Developments
39DP-PHINS, LUSBL performances
Using range measurements on suitable beacons leads to better positioning accuracy than USBL
At Least three ranges are required for LBL positioning
Station keeping with only two ranges is possible with INS and LBL combined (with INS initialization)
With multiple beacons and appropriate field geometry, DP-PHINS brings LUSBL feature to INS, taking full benefits of
Outliers rejection using PHINS high grade INSPrecision on ranges measurement (even with CW beacons)Error bounding by computed USBL positions
System accuracy is then directly related to the range measurement precision
40DP-PHINS, LUSBL Accuracy and Geometry
Accuracy of results is dependant on a combination of geometry and range measurement accuracy
If the beacons are “in line” with the vessel, LUSBL improvement will be
Optimum on beacon axisPoor orthogonally to beacon axis……. But still bounded by USBL precision
Accuracy will be optimum on both axis if there is a high “Angle of cut”
DP-PHINS naturally takes the best of USBL angles
and distances information
41DP-PHINS, LUSBL performances
In May this year iXBlue commissioned DP-PHINS onboard Subsea7 vessel Simar Esperanca.
Augmented-USBL brings x2 to x3 improvement over raw USBL, operating in over 1,350m on acoustics
LUSBL makes it even better Most of the USBL error is on the angle
measurements Range measurements are always consistent
Simply add a beacon... And take the full benefit of DP-PHINS !
DP-PHINS operating in 1364m (4475ft)
42DP-PHINS, LUSBL performances
The range measurements are far more precise than the angle measurements.
Slant range standard deviation on CW signals: <30cm 1DRMS
B14
B15
Vessel
2 beacons used during this testB14, 312m northB15, 285m south
LUSBL achievements in such a situation: Optimum precision on latitude No improvement on longitude
43
DP-PHINS, LUSBL performances
Performance improvement on latitude:
1DRMS Standard deviation on latitude
1DRMS precision (m) % slant range
USBL 5.09m 0.365%
DP-PHINS, USBL 1.65m 0.118%
DP-PHINS, LUSBL 0.32m 0.023%
At 1360m depth, LUSBL is 15.9x better than basic USBL5.1x better than augmented-USBL
Conclusions
45Conclusion: DP-PHINS as a PME
INS is a proven technology on Land, Under water, in Space, why not in DP?
INS has a long track record, Modern FOG based systems bring extreme robustness and reliability.
INS can produce heading and attitude data as well as positioning. INS Should not be aided just by GPS for DP applications.
DP-PHINS can make your acoustics as good as high accuracy GPS. The biggest benefits can be obtained by combining aiding sensors. With the addition of DVL, PHINS can even be considered a stand
alone PME for a significant period of time.
46Conclusion: Why moving to DP-PHINS ?
Intrinsic performance improvement due to PHINS IMU performances Augmented-USBL: x2 to x3 better than USBL with single USBL beacon Augmented-LUSBL: up to 16x times better with additional beacons
Sequential use of beacons for LUSBL ➯ battery savings of field transponders
Extended acceptable water depth for DP ➯ extended DP class Continued operation in case GPS outage ➯ recurring financial gain Easy refit of vessels with existing USBL
➯ unchanged USBL, pole… DP-PHINS can be installed anywhere on board
Fuel saving thanks to DP-PHINS output position smoothness Positioning system open to any additional sensor (DVL, deep water CVL…)
47
Thanks for your attention
www.ixblue.com