The Norröna: The Ocean and Climate

The Norröna System and Installation The water current measurements from the Norröna make use of  a device called an acoustic Doppler current profiler, or ADCP.  The ADCP's acoustic transducer is installed on the bottom of the ship and sends out four narrow acoustic beams down into the water and receives back the Doppler shifted signal reflected from plantonic particles and thermal inhomogenities in the water.  The Doppler shift of the returned signal is proportional to the water velocity along the beam relative to the acoustic transducer.  This data incombination with the ship's compass, GPS navigation data and a little bit of trigonometry, can convert these velocities into estimates of the absolute water movement in terms of its east, north and vertical components. The ADCP installed on the Norröna is a 75 kHz RD Instruments Ocean Surveyor (OS75) mounted in a void space between frames 105 and 106, about 60 meters from the bow (see figure below).  The  figure to the right (click on thumbnail image) shows a schematic of the entire ADCP system from the transducer inside its "seachest" to the deck unit and data acquisition system in one of the ship's electronics rooms on deck 8.  The "seachest" forms a water tight connection to the hull while the cofferdam provides a secondary containment.  A multi-conductor power and signal cable connects the transducer to the data acquisition and archiving system.  The primary heading data is supplied from a compass repeater tied to the ship's gyro compass.  Navigation information is supplied from a Thales ADU5 GPS receiver which provides very accurate heading information as well as the ship's position.  The data collection is controlled by the AutoADCP software which, in turn, controls and monitors RD Instruments' VMDAS data acquisition system.  When the vessel docks in Torshavn a two-way wireless internet connection to shore provides a means to control the system and down-load the data. The installation of the ADCP system on the Norröna presented some unexpected difficulties.  The initial installation took place in January 2006 at the Blohm and Voss shipyard in Hamburg, Germany.  We have installed and run ADCP systems on commercial and research vessels in the past but the Norröna presented a number of new challenges.  The first was that the ship is actually a full-scale floating hotel with all the amenities and finished accommodations that one would expect.  Leading several cables (see figure below) from the top of the ship to the bottom and from fore to aft meant that the panelling along seeming miles of corridors had to be carefully removed and then reinstalled.   And because we were dealing with a finished vessel, over a dozen watertight through-bulkhead cableways needed to be opened, the ADCP and other cables passed through, and then the pass-throughs re-cemented.

Cabling for the Norrona ADCP System

Bubble Sweep-Down Despite some surprises with the ADCP's initial installation, the ADCP was successfully installed during the one-week dry dock period.  It was during the shake-down cruise during the week following the relaunching of the ship that we encountered the real difficulties with the system although we were not sure what was happening for some time.  Various tests right after launch as well as later tests by an RD Instruments technician indicated that the instrument itself was working properly.  However the data return was extremely spotty and poor.  There were somewhat better results when the Norröna steamed through the fjords approaching Bergen, one of the ship's usual ports of call.  But as soon as the ship passed out of the fjords into the open sea the backscatter amplitude became irratic, the correlation level dropped to low levels (the OS75 is a broad-band instrument) and the percent good went to zero.  Occasionally, as in once every few minutes, we would get a possible profile but when the seas got rougher even those occasional profiles ceased.  Candidate sources of our problems with the data collection included internal machinery-generated vibration, propeller noise, electronic interference due to the long length of cable that necessarily ran along-side some of the ship's power cables, and the outside possibility that we were experiencing bubble sweep-down.  We talked to people experienced in ship vibration which suggested that there would be very little if any vibration energy in the ADCP's 75kHz frequency band even from the extremely noisy, to our ears, centrifugal oil separators located above and aft of the ADCP transducer.  Various tests with and without the ADCP connected and with various ship systems turned on and off finally showed that we were not getting any electronic interference.  That left bubble sweep-down as the only reasonable explanation for the poor rate of data return.  Bubble sweep-down, during which bubbles generated by the bow wave are swept under the hull, is something that the US Navy has dealt with but there does not seem to be much if any literature on how to deal with it.  The usual approach if bubble sweep-down is encountered is to mount acoustic devices on streamline pods below the ship.  On the relatively slow vessels of the research fleet, bubble sweep-down has sometimes been encountered in rough weather but is not usually a problem.  However in the case of the RV Iselin operated by the University of Miami, bubble sweep-down was shown with cameras mounted below the hull to be a significant problem.  This led to the installation of a simple pod for its acoustic sensors which generally corrected the problem.  The RV Thomas Thompson operated by the University of Washington has a permanent 19" deep pod in which all its acoustic sensors are mounted.  This has resulted in extremely good data from its ADCPs even in very rough weather.  In the past, bubble sweep-down has not appeared to be a significant issue in the voluntary observing ship programs that make use of commercial container vessels and cruise ships.  Our own experience with the MV Oleander, a relatively small (~400') container vessel operated by Bermuda Conatiner Lines that makes weekly runs between New York and Bermuda traveling at about 17kts, has indicated that when the ship is loaded during the Bermuda-bound legs that the data return is usually quite acceptable.  On the return New York bound leg of the trip when the ship is lightly loaded, draws ~1m less and the bow rides higher, it takes reasonably good weather for good data return.  The hull of container vessels, and this includes the Oleander, are generally fairly blunt and generate a significant bow wave which one would think would lead to major bubble sweep-down problems.  But that does not appear to be the case.  The Norröna runs only somewhat faster than the Oleander at 21kts.  The hull, on the other hand, looks similar to that of high speed cruise ships with a sharp entry which one would normally expect to generate a minimal bow wave and thus, minimal bubble sweep-down.  In addition, the Norröna is almost always ballasted to maintain a constant draft, whether it is empty or full.  One potential source of the bubbles is that the Norröna has two enormous bow-thrusters located just below the waterline (see the photo below) which could draw in bubbles when the ship pitches.  Another possiblity brought to our attention after consultations with naval architects familiar with high speed ferry design and their bulbous bows is that the sharp intersection between the bulbous bbow and the hull creates a potential pathway for bubbles to get under the hull. Evolution of the Bubble Fairing With the painful conclusion that bubbles were all but completely blanking out the ADCP we were left with two choices.  Either remove the ADCP and give up on the scientifically ideal route between Europe and Iceland, or try to come up with a viable solution.  We chose to try to save the installation and the route by investigating the feasibility of installing a fairing.  And rather to our surprise, Smyril Line was willing to go along with this approach.  For those who are unfamiliar with commercail ship design, virtually all commercial vessels are falt-bottomed over most of the hull with no keel projecting down.  Thus, any fairing we might install would be the deepest and most exposed portion of the hull. The first item of business was to find out whether a fairing would do any good at all.  Luckily for us, the Norröna was scheduled for a dry dock period in May 2006 which afforded us an opportunity to attach a temporary fairing.  By temporary we mean that it was expected that at some point the fairing would fall off either through finally yielding to the flow past the device or because the unit was bumped while maneuvring in the shallow waters of the ports of call.   Given the lack of any references on fairing design the sizing of the fairing was a somewhat arbitrary exercise.   We made the assumption that the bubbles would be contained in the turbulent boundary layer running along the bottom.  The boundary layer thickness was estimated to be about 15 cm thick at the location of the transducer so a fairing about 20 cm tall with a suitable lip should penetrate the boundary layer and push the bubbles aside.  The free-flooding temporary fairing was foil shaped, 2.75 m long x 0.81 m wide x 0.20 m tall, made from 1/8" stainless steel, filled with expandable foam and covered with a 3/8" thick Lexan sheet.  At ~20 cm from the transducer the lexan sheet was acoustically transparent to the ADCP.  The drawing and photos below show the fairing.  In the first photo the bottom of the fairing is facing up and the conical cutout for the acoustic beams is visible.  The fairing was glued to the ship using Dow Corning 5200 marine cement.  The temporary fairing stayed attached to the ship for about 10 days and during that time we were able to collect data as the ship made a trip from Torshavn to Europe and almost all the way back.  The percent good from that run is shown below indicating that we were getting nearly 100% good profiles down to more than 400m in the Faroes-Shetland Channel.  The fairing seems to have fallen off just before the ship reached Lerwick in the Shetland Islands.  The contrast with and without the fairing is obvious and the value of a bubble fairing was demonstrated.  But gluing a fairing to the hull was not going to be a permanent solution. As an aside, when we finally got some data to look at it was clear that the ADCP itself may have been defective. Shortly after the temporary fairing was installed in May 2006, we discovered that the ADCP instrument had developed a malfunction, probably shortly after the first two weeks of operation. With the temporary fairing the ADCP appeared to work  but the velocities calculated from the Doppler shifts did not seem plausible. After extensive attempts at troubleshooting with RDIntruments technicians, we first received a replacement deck unit as a loan from RDI, but this did not help. The problem appeared that the problem was with the transducer.  So when the vessel was drydocked in the 3rd week of January 2007 for work on the ship's stabilizers, the transducer was removed and returned to RDInstruments for analysis and repair.  The repaired transducer was reinstalled when the vessel was drydocked at the end of October 2007. With the evidence that a fairing might do the job, Smyril Lines was amenable to the idea of a permanent, non-frangible fairing provided that it passed inspection by Det Norska Veritas (DNV), the Norwegian ship certification agency.   Bay Marine Inc of Barrington, RI undertook the task of designing the fairing and submitting the drawings to DNV.  The new fairing differed from the temporary fairing in that it was slightly narrower, the lip around the bottom was thicker, and by the presences on three rugged bumpers installed ahead of the fairing.  The bumpers were intended to ward off debris that might hit the fairing.  The drawings were OK'd in time for another dry dock period in late October, 2007.  While the idea of a bubble fairing was being pursued, a second method that would acoustically detect the presence or absence of bubbles under the hull was being investigated.  On a rather short time frame for such a device, a protype 250kHz transducer and data logging system was developed in time to run some tests during the week before the Norröna was scheduled for the October dry dock period.  These test were rather uncertain but seemed to indicate that there might be a lower concentration of bubbles out board of the transducer's initial location.  With some trepidation, the decision was made to move the transducer and install the permanent fairing about 3.2m outboard of the original position.  The photo below shows the fairing from the front during installation and shows the rather substantial bumpers.  Note also that the leading edge of the lip around the fairing was angular and not rounded. Although we had great expectations with the new fairing, it did not live up to our hopes and data return was still rather poor.  Better than before, but still poor.  This led to further assessment of the problem and more consultation with experts in  the fluid dynamics around ship hulls.  There were a number of possibilities under consideration, 1) that we had moved the transducer into a region with a greater bubble concentration and/or a thicker bubble layer,  2)  that the bumpers had altered the flow such that vortices rolled the bubble layer underneath the fairing, or 3) that the sharp leading edge was forming cavitation bubbles that then flowed passed the transducer.  In November 2007 the ship encountered some unusually bad weather which caused one of the stabilizer fins to break off and punch a small hole in the side of the ship.  This led to another dry dock period in February 2008 and we took that opportunity to make some relatively minor modifications to the fairing.   During this dry dock period we removed all three of the bumpers, added two inches to the lip of the fairing and carefully rounded the leading edge.  The modified bubble fairing is shown in the photo below showing a much smoother and less obstructed structure.  Results with this version of the fairing were somewhat  better and we have gathered a reasonable amount of data between the Faroes and the mainland.  Some of this improvement might simply be that the first few months with the modified fairing have been during the late spring and summer when the winds and waves are less.  Comparing the returns with the weather between the Faroes and Norway certainly suggests that the returns are better when the winds and waves are low, and less good under more adverse conditions.  Also, we're still getting poor returns over the sections the run between the Faroes and Iceland which clearly  traverses an important  limb of the North Atlantic current into the Norwegian and Greenland Seas.

Norröna's bulbous bow and bow thrusters	Temporary bubble fairing	Temporary bubble fairing	Temporary bubble fairing
% Good with temporary fairing	1st permanent bubble fairing	Permanent bubble fairing	Modified bubble fairing

The updated version of the thru-hull bubble sonar operates at 200 kHz as before but with an improved digitization rate of 24 kHz, decimated to a sample rate of 240 Hz, yielding a vertical resolution under the hull of 3 mm.  The operating frequency is determined by the acoustic impedance of the 12.5 mm thick steel hull such that the hull is nearly acoustically transparent and there is maximum signal penetrating into the water.  The transducer was moved to 18 places in the void space where the ADCP could be located to see if there was an optimal location.  At each location 100 profiles were obtained over a period of one minute returning 120 backscatter estimates between 104 and 600 μsec after the ping.  The results that were obtained by the sonar (see below) indicated high backscatter close to the hull that decayed to background at between 10 and 20 cm.  The background values corresponded closely with those obtained when the ship was still at the dock.  Occasionally, we got high backscatter from a few pings farther from the hull with the farthest high backscatter return from about 40 cm.  The weather during the sonar runs was fairly benign by North Atlantic standards, with 10 to 18 kts of wind and average wave heights of 8 to 11 ft as reported by the British meteorological buoy 64046.  So the results do not reflect the severe bubble conditions that typify winter seas but they do suggest that there may be a “usual” bubble cloud close to the hull of some 10 to 20 cm thick and that deeper bubbles are also possible but the sonar data are not sufficient to characterize those more extensive bubble clouds.

Results from the bubble sonar showing 100 backscatter profiles at a number of locations as a function of distance below the hull.

The underwater camera was an attempt to get unequivocal evidence about the character of the bubbles under the hull.  The difficulty with this approach was to find a camera that could operate autonomously, record the pictures internally and be diver deployable.  It turns out that we were not the first with requirements for an autonomous underwater video camera.  Greg Marshall’s group at the National Geographic Society have been developing their Crittercam for several years so that they could attach it to various sea animals and record their behavior.  The latest version of the Crittercam is remarkably compact and can record up to eight hours of video (and audio) data on internal solid state memory.  Marshall graciously allowed us to use one of his cameras and Eric Bergenpas set it up so that it could record one minute of video data every four hours for up to five days.  The next step was to develop a method to attach the camera to the hull in such a way that a diver could deploy it at a number of locations while remaining attached at the Norrona’s usual cruising speed of 20 kts.  That was accomplished using a series of permanent magnets that together developed a grip of about 1600 lbs together with a lever and cam arrangement that allowed the unit to be detached from the hull.  Then the whole camera and clamp were housed in streamlined fairing to minimize drag.  The magnetic clamp and fairing were tested on the high speed Port Jefferson to Bridgeport ferry with the assistance of Mark Wiggins and graduate student divers Brooke Rogers and Amber Stubler.

Streamline fiberglass shell, and Crittercam on magnetic clamp

Assembled underwater camera system

Camera positions under void space

The camera was deployed at five different locations with the help of Ebba Mortensen of the Faroese Fisheries Institute and Edvard Kjeld a professional diver operating in the Faores, to study the character of the bubble clouds and to get some idea of their spatial distribution.  That there are bubbles under the ship is indisputable from the camera results and they are clearly the cause of the poor data return from the ADCP.  The most informative results came from the videos taken during daylight hours when light from the surface illuminated bubble clouds from the side.  In the figure below showing the bubble fairing from the side, one can see the bubble cloud approaching from the right.  The fairing is 21 cm high indicating that this particular bubble cloud is roughly 30 cm deep.  This video was obtained when the winds and sea state were relatively mild.  When the conditions are much rougher, the camera’s vision is often obscured by the bubbles next to the lens so our ideas about what is happening may be biased.  The second figure below was taken looking forward at a location closer to the centerline of the ship.  The grey clouds visible below the hull are the bubble clouds approaching the camera.  A few frames later the camera is obscured due to the bubbles hitting the lens.  While the bubble clouds are undoubtedly produced in the turbulent bow wave as the ship pitches up and down, the shape of the clouds seem quite steady for the two or three seconds that they show up in the videos.  It is also clear that the larger bubble clouds are produced by the pitching of the ship as they come at the camera at fairing regular intervals.

Camera view looking at the fairing	Camera view looking forward

Clicking on the following links will bring you the most illustrative videos for the first four deployments: #1, #2, #3, and #4.  The positions of the camera for each of the videos is shown in the figure above right.  In the first video the camera is upside down so up is down and left is right.  Overall, the combined camera and sonar result together with the results from the first temporary bubble fairing suggested that there will be fewer bubble problems if we moved the ADCP closer to the centerline of the ship.

The last part of the investigation involved computational fluid dynamics simulations performed by Bob Fratantonio, a graduate student in ocean engineering at URI, using Floworks software.  The questions we wanted to resolve were: 1) was there a better overall shape to the fairing than the initial one, in particular would a more pointed fairing reduce the stagnation pressure at the leading edge of the fairing, 2) would extending the lip forward to 8 cm (from the current 4 cm)  help reduce the boundary layer flow under the fairing, and 3) would vertical fins or chines placed ahead of the fairing cause sufficient upwelling under the hull to bring bubble free water up against the fairing and transducer.  Each of these questions was addressed in succession and then finally in combination to see how the overall system would work. 

The results of altering the shape of the fairing to be more pointed does reduce the stagnation pressure at the nose of the fairing slightly and produces less of a downward deflection of particles flowing along the centerline as shown in the figure below.  The figue shows path of particles released ~0.5m ahead of the fairing and 0.2m below the hull.  The reduction of the stagnation pressure on either side of the centerline is more substantial with a greater effect on the flow field.

The present lip on the bubble fairing sticks out about 2 in (5 cm) with the hope that it would reduce the tendency for water and bubbles to flow under the fairing.  This seems to work to some extent but we were interested to see whether increasing the lip to 4 in (10 cm) would improve its performance.  The figures below show a centerline cut of the vertical velocities using both vectors and color.  It appears that the downward vertical velocity with the extended lip is less, but a greater effect shows up in the vertical velocity 5 cm under the fairing.  There, the extended lip produces vortices on either side which produces an upwelling toward the face of the fairing which seems like it would be beneficial by bringing deeper water toward the transducer.

Z-Velocity Side Cut Plot - 2 Inch Lip	Z-Velocity Side Cut Plot - 4 Inch Lip
Z-Velocity 5cm from Fairing Face – 2 Inch Lip	Z-Velocity 5cm from Fairing Face – 4 Inch Lip

The last part of the investigation involved a study of whether vertical fins oriented somewhat across the flow could be designed in such a way that they would cause water from farther below the ship, and hopefully with less bubbles, to upwell toward the hull and sweep away the bubble laden waters.  A number of configurations were tried at various distances ahead of the fairing.  The basic idea was that a pair of fins, spreading out from the extended center of the fairing would generate a pair of counter-rotating vortices and upwelling along the centerline.  After some experimentation, a pair of fins 20 cm tall and about 2 m long in the shape of a truncated hyperbolic tangent worked the best.  In the figure below chines are straight fins while rails are fins in the shape of a hyperbolic tangent.  The top of the plot represents the bottom of the ship and leading edge of the fairing body is at 0m.  Clearly the most effective configuration makes use of the hyperbolic tangent rails 10 m ahead of the fairing which is able to draw water from 0.5 m below the ship and pull it to less than the height of the fairing itself, which is 20 cm deep.

Particle trajectories along the centerline for a variety of fairing shapes, fin shapes and distances ahead of the fairing.

In order to extend the width of the upwelling region, the rails extend out to 0.75 m on either side of the centerline (the fairing itself is about half this width) so that the upwelling covers the essentially the whole fairing as shown in the next figure.

Plot of the vertical velocity about 25 cm below the hull of the ship.  Blue is up toward the hull and yellow is downward

The Norröna entered the drydock at the Blohm and Voss repair yard in Hamburg, Germany in January, 2009.  At that time the existing fairing and coffer dam were removed and a more pointed fairing with a 10 cm lip and leading hyperbolic tangent rails was constructed.  These will be installed about 0.8 m inboard of the original ADCP location and 2.8 m from the keel, in the same void space as shown below.  The curved 0.2m tall rails were installed in a void space about 10m forward of the ADCP.  The pictures below show the new fairing under construction and the new installed and painted fairing and rails.  The ship was refloated by the last week in January, 2009 and we started to get the first results with the new arrangement by the mid-February.  It is clear that the new system works better than any of the previous interations but we still unable to get reliable data on the Iceland to Faroes leg of the trip.  However, the data collection between the Faroes and Denmark across the climatically important Shetland/Faroes Channel, especially on the eastbound leg of the trip, has reached acceptible levels and we are encouraged that we are headed inthe right direction.

Results While we have been having difficulties in collecting velocity data, the ADCP, data collection and automated control of the system has operated very well.  At least once a week we check on the system's status remotely and download the recent data files.  The map below left shows all the ship tracks from November 2007 through December 2008.  As mentioned in the introduction, the Norrona's weekly track spans the distance from Seydisfjordur in Iceland to one of three ports in Denmark with intermediate stops in Torshavn,  Bergen and occasionally Scotland and Lerwick in the Shetlands.  In the winter months there is less demand for the Iceland portion of the route so the Norrona's are restricted to the eastern half of the range.  Even though the velocity data coverage has been sporadic, the ADCP does collect the near surface temperature data which is also downloaded once a week.  A time-distance plot of the surface temperature is shown in the figure below right .  Click on the figure to blow it up for better viewing.  The green tracks in the map show the location of the temperature data included in the seasonal temperature evolution plot to the right.  Among other items visible in the plot is the southward flowing cold water along the eastern Iceland coast, the slightly cooler water flowing northward along the Faroes' west coast, the abrupt difference in seasonal warming between the deep Faroes-Shetland Channel and the shallower shelf of the North Sea, and the relative delay in maximal summer temperatures between the North Sea shelf waters and those ove the Iceland-Faroes Ridge.  These results from the continuous temperature sensor emphasize the power of the Norrona's route for monitoring this complex portion of the ocean and suggest that the velocity we hope to be collecting with the repositioned ADCP and fairing will be of immense value.  With the February 2008/February 2009 modifications to the bubble fairing, velocity data collection has improved to the point that we fairly consistent sections with ADCP data across the Shetland Channel and quite a bit more in the North Sea. What stands out from these first sections is the high level of eddy activity in the Shetland Channel (actually everywhere!). Even the Shetland (or Slope) Current varies significantly. Based on the peak speed from the nine sections so far in the vicinity of the Slope Current, the peak speed is about 50 cm s-1 with a 25% standard deviation, with considerable variation in position (ie. not tightly locked to the slope) and direction. We do know from recent Lagrangian work that this area is very ‘eddy-active’ and it would appear that this activity impinges upon the Slope Current as well. On the eastern side of the Channel, the flow is always to the north as one would expect, on the Faroe Slope side the flow is generally to the south, but here reversals are quite common and these may well be due to the semidiurnal tide, which appears to be very strong on the Faroe Plateau (Davis et al., JPO, 2000). In short, in the center of the channel the velocity variability is eddy-dominated, on the Faroe shelf it is probably tidally dominated, and on the Shetland Slope we are less certain, both eddy and tidal currents contribute to the variability of the Shetland Current.