Design Methodology and Numerical Analysis of a Cable Ferry

Transcription

1 Dean M. 1 (M), Ryan S. Nicoll 1 (V), Tony Thompson 2 (M), Bruce Paterson 3 (M) 1. Dynamic Systems Analysis Ltd., Dallas Road, Victoria, BC, Canada 2. EYE Marine, Suite 1, 327 Prince Albert Road, Dartmouth, NS, Canada 3. British Columbia Ferry Services, Inc., Blanshard Street, Victoria, BC, Canada Cable ferry systems offer many advantages including simplicity, low maintenance, low operations risk, and lower energy use when compared to freely maneuvering ferries. In this paper, a proposed ferry system on a long 2000m crossing is assessed using advanced time domain analysis with verification by model scale tests. The results are used to ensure a safe and reliable design with similar level of service to the incumbent ferry system on the existing route. KEY WORDS: Cable ferry; passenger vehicle; roll-on/rolloff; hydrodynamics; mooring; dynamic analysis; time domain simulation INTRODUCTION Cabled ferry systems have been in use for centuries. They are used all over the world to due to their simplicity and frequent advantages over freely maneuvering ferries. Cable ferries are connected to both shores of a crossing with a cable or cables that facilitate ferry transit. A modern vehicle and passenger carrying cable ferry is typically connected using wire ropes and propelled by a winch onboard the ferry. Cable ferry routes typically vary from 100 to 1500 m with the most common range from 800 to 1000 m. Figure 1 shows the Henry Nase cable ferry operating in New Brunswick on a 615 m crossing at Gondola Point. Aprons on either end of the ferry are used for loading and unloading of vehicles on each end of the crossing. Figure 2 shows a cable ferry berthing. The figure indicates the cable which the ferry uses to guide and pull itself through the crossing. The cable transit shown here is 615m and is located at Gondola Point. Cable ferries are used for a number of reasons. They are simple structurally due to their flat bottomed barge shape and require relatively simple mechanical systems. The mechanical systems are minimized by the use of a winch drive system, replacing rudders and propellers. In addition, fuel consumption is low because the propulsive efficiency of the winch drive system is greater than a propeller. The reduced complexity requires a smaller crew and lower operational and maintenance requirements results in reduced operational costs. Lastly, the cables guide the ferry and so navigation and complex berthing operations are simplified, which reduce operational risks. Most cable ferries operate on shorter routes in protected coastal areas or on rivers or lakes and therefore can be designed by marine engineers and naval architects based on common practice, experience, and standard analysis methods. However, longer crossings require additional consideration. The longer the 294

2 crossing, the more severe the potential environmental loads due to increased exposure and fetch, which in turn can result in encountering higher wind speeds and wave conditions. As a result, the motion of the ferry and resulting stresses on the cable system must be considered to ensure safety and reliability, and to achieve an acceptable level of service specified by the operator throughout the lifetime of the ferry. Figure 1: Henry Nase Ferry operating in New Brunswick, Canada; crossing distance 615 m Figure 3: Buckley Bay to Denman Island ferry route; crossing length 2000 m CABLE FERRY OPERATIONS Figure 2: Cable ferry docking with traction cable indicated British Columbia Ferry Services, Inc., (BC Ferries) is presently engaged in the replacement of a ferry offering service between Buckley Bay on Vancouver Island and Denman Island, on Canada s west coast. The existing ferry operates on a straight route, approximately 2000 m in length, in a relatively wellprotected area, as shown in Figure 3. Because of the aforementioned benefits, a cable ferry has been proposed for use on this route. However, when designing such a system, the cables have a significant impact on many aspects of the design of the vessel, and many additional considerations arise. For example, relative motion between the ferry and the docking structure are significantly affected by cables. The cable fatigue life must be considered, and ferry seakeeping behavior is affected by the cables. This paper reviews work directed by BC Ferries to assess the design. An emphasis is placed on the numerical analysis work completed by Dynamic Systems Analysis Ltd. (DSA) and E.Y.E. Marine Consultants (EYE). This work included assessment of level of service, cable loads, docking loads, and ferry motions for the proposed ferry. The unique aspects of the project are emphasized. Ferries are used around the world as a simple and efficient system where traffic demands will not warrant the construction of a bridge or where a fixed structure is not practical or would impair the passage of other marine traffic. The cable ferry is limited by applications where access roads and terminals can be arranged in a direct line to facilitate loading and unloading. Since the ferry follows the track of the fixed cables and there are no navigating requirements other than docking and maintaining watch, the crew can often be reduced consistent with safety requirements to further reduce costs. Most of the cable ferries have evolved to using a single drive cable on the centerline of the ferry. This cable is anchored at both ends by substantial ground anchors with allowable pull greater than the breaking load of the cable. The cable is lead through fairleads on either end of the ferry and around a pair of bullwheels. A rendering of a typical hydraulically driven bull wheel system is shown in Figure 4. At least one of these bullwheels is powered, however both can be powered if redundancy or additional pull is required. Cable ferries can use additional guide cables to control drift or to increase the level of safety in the event of a cable break. Depending on the traction load required, more than one drive cable may be necessary. Most cable ferries utilize a hydraulic drive motor directly coupled to the shaft which can deliver the required torque at a speed of about 50 RPM. There are electric drives that are battery operated as well as mechanical drive systems. The hydraulic drive is relatively efficient with an estimated drive efficiency of about 80% for a well matched pump and motor. It is this high efficiency that makes the system attractive to operators as fuel costs are substantially reduced. A 42 m 24 car ferry can reliably maintain a speed of 7 knots with a 50 kw hydraulic motor in normal full loaded operating conditions. 295

3 The lateral drift of the ferry from the direct route is caused by wind and current loads. The drift is controlled by the tension on the cables. It is generally necessary to pretension the cable to reduce the drift and to provide enough friction on the drive bullwheels to prevent slippage. Pretension may also be required to guide the ferry when docking at a terminal. In general, the pretension of 1/5th of the breaking load is not to be exceeded. It is not necessary to lift the cables off the bottom, although this occurs over deeper sections of the crossing. When the ferry is subjected to an external lateral force the cable catenary is reduced and naturally increases the tension in the cables until a balance of environmental forces against resultant cable tension is achieved and the system is in equilibrium. Figure 4: Rendered model of a typical hydraulically powered traction winch drive system complete with bullwheels Analysis of the cable ferry system by equalizing of the resultant forces is a quasi static approach. The resulting forces have a safety factor of about 3 applied to compensate for dynamics and wear that are not considered in quasi static methods. An important control in using quasi-static methods is to correlate the stretch of the cable to the predicted drift and catenary shape to ensure that the stress/strain relationship remains correct. While the quasi-static methods have been used successfully on many ferries in service with great success, the method should not be relied upon when dynamic conditions become significant. On the subject ferry it was determined that combination of a large ferry with a longer transit and potentially higher transverse winds warranted the use of dynamic methods to accurately determine the cable sizes and forces. accommodations and machinery spaces and supports a raised control station (bridge) providing good visibility to the docking areas and the car deck. The central deckhouse facilitates fast and safe loading and unloading by maintaining straight lanes in conjunction with wide loading ramps with dual lane loading/unloading. Figure 5: Profile of concept ferry Table 1: Proposed ferry physical properties and characteristics Length OA 80 m Beam (WL) Depth Draft Mass Roll radius of gyration Yaw radius of gyration Pitch radius of gyration 16 m 2.1 m m 526,000 kg 6.41 m 19.6 m 19.6 m Vehicle capacity (AEQ = Automobile 50 AEQ Equivalent unit, 6.5m) Complement 150 Vessel life 40 years The proposed ferry is propelled by a bull wheel around which the traction cable wraps. In the proposed design, all machinery is located above the main deck for easy and safe access. Two additional cables on either side of the ferry provide additional guidance during the crossing, as shown Figure 6. Figure 19 (at end of paper) shows how the cables interact with the seabed during the crossing. The pretension in the guide and traction cables for this application will be around 200 kn, although it is possible to operate at pretensions near 100 kn as well. A inch 6x19 IWRC wire rope, with a mass of ~7.0 kg/m, will be used for the guide and traction cables. CONCEPT OVERVIEW The cable ferry concept shown in Figure 5 has been proposed for use between Vancouver Island and Denman Island. Key parameters for the proposed ferry are provided in Table 1. The route length is approximately 2000 m. The ferry will carry approximately 50 vehicles or 50 Automobile Equivalent (AEQ) units on a roll-on / roll-off deck. The ferry is about 80 m long, with a centerline deckhouse that contains passenger Figure 6: Cable ferry general arrangeme: 296

4 The bull wheel must have sufficient power to pull the ferry through the crossing at the required transit speed, which may be slightly slower than a conventional ferry as there is less time lost in docking. For the project under consideration, the ferry is expected to have a service speed of 7.5 knots with 300kW power available for propulsion. The ferry must also have sufficient power to pull itself into the dock during heavy weather conditions. The ferry will berth with the terminal on a floating pontoon that is connected to a hinged ramp. The floating pontoon structure, held laterally and in yaw by piles. The docking system schematic, shown in Figure 7 (piles not shown), illustrates how the cables pass through the pontoon to shore where they are anchored. Figure 20 shows the ferry next to the pontoon, and the passage of the cable through the pontoon and ferry. Figure 7: Ramp and pontoon general arrangement CONECEPT VALIDATION Prior to commencing detailed dynamic analysis of the ferry, a validation of the ferry concept was completed. The selection process consisted of checking weights, intact stability, and damaged stability against regulatory standards. The approximate weights of the cables were included in the lightship weight. A preliminary bulkhead arrangement was determined to confirm that the floodable length standards were met. Structural calculations were performed to confirm the weights and longitudinal strength were acceptable. Finally a preliminary resistance calculation was prepared to identify that service demands and speed requirements would be met. A hull optimization process was carried out that included a preliminary evaluation of the ferry motions in waves and at speed using a strip theory analysis. The vessel s docking system, as shown in Figure 7, facilitates a non-flat bottomed ferry to be used. This is in contrast to the docking system shown in Figure 2, which relies on a flat bottom. Since a non-flat-bottomed ferry may be used, the vessel could be designed to incorporate minimal deadrise to improve the longitudinal directional stability of the barge type hull (addition of appendages such as bilge keels were considered). Substantial overhangs to minimize deck wetness were incorporated and consideration of the guide cables and traction cable propulsion system were incorporated. All of the validation work confirmed that the ferry exceeded regulatory standards and would result in a safe and seaworthy vessel. DYNAMIC ANALYSIS REQUIREMENTS To comprehensively assess the proposed cable ferry system and its associated structures requires analyzing the interplay between the cables, the ferry, the shore works (i.e. pontoon, ramp, piles), and the mechanical systems on board the ferry. The global dynamics of this system must be understood to accurately assess the loads and behavior of the ferry during transit and berthing. BC Ferries initiated a project to determine loads and the behavior of the cable ferry under a range of environmental conditions through dynamic analysis. This analysis was used to support the detailed design of the ferry, the shore structure, and allow for specification of system components (wire rope size, bull winch power capacity, etc). The dynamic analysis of a cable ferry is unique in several respects. The ferry can be considered both a moored platform and a moving vessel, so consideration to its motion must take into account coupled vessel and cable dynamics. The assessment of the station-keeping behavior of moored platforms is well established. Experience, standards, and methodologies exist which provide guidelines for their analysis. Common mooring standards include: API RP 2SK Recommended Practice for Design and Analysis of Stationkeeping Systems for Floating Structures ; DNV OS-E301 Position Mooring ; and; ISO Stationkeeping systems for floating offshore structures and mobile offshore units. However, these standards provide analysis methodologies focused on offshore platforms and not coastal marine applications. Assessment of the dynamic behavior of barge shaped vessels is likewise well established in codes and practice, but the design techniques used for these vessels do not typically include analysis of the vessels attached to cables. Thus analysis of a cable ferry must draw upon aspects of both of these fields of practice to ensure accurate analysis and achieve acceptable risk levels for operators. Regardless of the approach or technique used, whether numerical or through physical model tests, the dynamic analysis must allow for determination of: i. Vessel / pontoon impact loads ii. Relative motion of the pontoon and ferry while docked iii. Loads on shore structures when ferry is docked iv. Loads in cables (safety factors) v. Loads on cable anchor points vi. Powering requirements for ferry bull wheel system vii. Braking load requirements viii. Excursion from transit centerline during extreme conditions ix. Cable fatigue life x. Operability given expected environmental conditions at the site xi. Ferry motions (accelerations and roll) during transit and docking xii. Loads on cable fairleads on ferry xiii. Loads on cables in damaged condition (e.g. one cable has been disconnected) xiv. Assessing modal frequencies of cable/ferry with respect to environmental loads xv. Assessing the effect of modifying cable pretensions on maximum loads and excursion from transit centerline 297

5 METHOD OVERVIEW For the project, the nonlinear time domain simulation software, ProteusDS, was used by DSA to simulate the coupled ferrycable system dynamics. The software is used for coupled analysis of offshore structures, moorings, and a variety of ocean equipment and technologies (, 2008; Nicoll, 2011; Nicoll, 2012). It was selected for use for several reasons. First, it can model the cable loads using a nonlinear finite element cablebeam model, which can incorporate axial, bending, torsional, and viscous drag effects. Second, it incorporates bathymetric variation of the crossing and the resulting change in cable loads during interaction with the seabed from friction and contact forces. Third, the software can model rigid body floating vessels along with winch controller systems. Lastly, the software allows for modeling and assessment of ferry-dock rigid body interaction, including cable and contact loads. Ferry Hydrodynamic Model The seakeeping analysis software ShipMo3D was used to generate a hull-specific hydrodynamic database to incorporate the ferry hydrodynamic effects in the analysis (McTaggart 2012; McTaggart 2010). The hydrodynamic database allows the computation of wave radiation, diffraction, and excitation loads as a function of forward speed of the ferry in the time domain. The software uses a panel method approach in conjunction with potential flow theory to determine wave excitation, radiation, and diffraction forces on a ship hull. This data was generated for the ferry hull using a panel method code instead of a strip theory approach due to the relatively shallow draft and low length-tobeam ratio; comparison of results between a strip theory program and ShipMo3D with model testing results confirmed that the strip theory approach under-predicts the maximum roll angles. The hull mesh, with panel size limited to 0.2 m 2 is shown in Figure 8. This fine mesh was found to be required due to the shallow draft (0.61m) which led to large irregular frequencies in the potential flow solution. Removal of the irregular frequencies was accomplished by decreasing the panel size in the mesh. Figure 8: Ferry hull mesh as shown in ShipMo3D Wind Model In fetch-limited coastal regions, wind loads are an important source of loading. A model of the ferry topside was created for use in the time domain simulations. The topside geometry is roughly represented, and a drag coefficient due to wind velocity in the principal degrees of freedom (surge, sway, heave) is used. Figure 9 shows the topside geometry used in the simulations. Wind loading is applied based on the relative velocity at each discrete panel in the mesh. A wind shear profile was used with lower wind speed at the water surface. Key ferry simulation properties are given in Table 2. Figure 9: Ferry topside wind loading model Table 2 Ferry simulation parameters Parameter name Value Mass 526 x 10 3 kg Roll radius of gyration 6.41 m Pitch radius of gyration 19.6 m Yaw radius of gyration 19.6 m Vertical location of center of m gravity from baseline Superstructure surge wind drag 1.15 coefficient Superstructure sway wind drag 0.97 coefficient Draft 0.61 m Finite-Element Cable Model The cables were represented using a cubic finite element cable model. The mechanical properties given in Error! Reference source not found. were calculated from data provided by the manufacturer, and recommended values for axial rigidity for stranded wire rope from DNV (DNV, 2010). Bending and torsional stiffness values for the cable were specified; however, the axial rigidity, drag, and weight are the most critical parameters for the dynamic simulation. The contact forces between the seabed and cable are computed using a seabed stiffness and damping coefficient and cable penetration into the soil. For all simulations, the cables were connected to the ferry via a tangentially sliding connection at the bow and stern of the ferry, as shown in Figure 10. The sliding connection represents the swivel sheaves on the ferry. The connection type use stiffness and damping to keep the cable laterally aligned to the connection point but does not provide any resistance to relative motion in the cable tangential direction at the connection point. Sliding friction in the fairleads is not significant and was neglected. The sliding connections are the primary means of transmitting cable reaction loads to the ferry. 298

6 Table 3 Cable properties for simulation (final cable selected had a diameter of m and a mass per unit length of 7 kg/m) Parameter name Value Diameter m Mass per unit length 5.92 kg/m Axial rigidity/ea x 10 9 N Flexural rigidity/ei 8.2 knm^2 Torsional rigidity/gj 8.2 knm^2 Normal drag coefficient 1 Tangential drag coefficient 0.01 Normal added mass coefficient 1 ferry. The traction winch which increases the tension in the traction cable to propel the ferry is indicated. During transit simulations, the dock structure was neglected and the cables were pinned to the static pontoon dock connection location. During crossing, the pontoon is too far away to significantly influence the ferry motion and the reduced complexity increased execution speeds of the time domain model. For docking simulations, the cables were pinned to locations on shore that correspond to shore anchor locations. In addition, several sliding connections were used to constrain the guide and traction cables to guide the lines along the pontoon before termination at the shore anchors as illustrated in Figure 11. Winch Model In order to stop or drive the ferry forward, a thrust force is required. A PID (proportional-integral-derivative) controller that observed the forward speed of the ferry and applied a thrust force on the ferry with equal and opposite reaction force on the traction cable to adjust the speed accordingly was used. The controller used integral control exclusively to ensure an accurate steady state forward speed. This automatic controller ensured the ferry reached a relatively steady desired forward speed value during transit in spite of environmental loading conditions. The applied force was limited knowing the winch power and load capacity. The power was limited to 300kW in the simulations using the controller. This value was selected based on previous experience and resistance calculations. The actual traction and braking power needed during a given simulation was computed through a post-processing step after a simulation was completed. The braking or traction power required to stop or drive the ferry is the thrust force produced by the controller multiplied by the forward speed of the ferry. Simulation Measurements In the time-domain software, virtual instruments were placed at specific locations on the ferry model to measure key pieces of information throughout the numerical simulation; the key data was written to text files for post-processing. For example, acceleration probes were placed at several points on the deck at the bow, stern, and passenger lounge near the center of the ferry to record linear acceleration seen at each point through the simulation. This was done because the bow and stern will be the location of maximum heave acceleration on the ferry due to ferry roll and pitch effects. Figure 10: Time domain model setup schematic showing placement of sliding connections on the ferry which model the swivel sheaves where the cable loads are transferred to the cable In addition, air gap probes were placed at several locations around the bottom of the hull of the ferry. The air gap probes are placed at various locations on the hull surface location underwater and indicate the distance between the sea surface and the probe location. The air gap probes are used to give an indication on whether large portions of the hull are exposed to air due to shorter length waves in storm conditions. This is important for structural reasons in addition to assessing the validity of the numerical model. 299

7 Dock Model To examine the interaction of the ferry with the dock, further time domain simulations were completed. The dock model was represented by an articulated body, or a serial concatenation of rigid bodies connected by joints. The dolphins were held fixed rigidly against the seafloor. The floating pontoon was connected to the dolphins by a joint that allowed heave, pitch, and roll. The ramp was neglected but the mass of the ramp was lumped in with the floating pontoon. The general setup of the model is shown in Figure 11 and Figure 20. load with relative heading to the ferry between steady state simulation and the wind tunnel test data can be seen in Figure 13; the maximum wind load of approximately of 108kN was seen in the sway direction at a wind speed of 55 knots. It was assumed that the general layout and geometry of the wind tunnel model matched the configuration used in the time domain model. The surge and sway drag coefficients used for the time domain model were determined by using the same wind velocity and shear profile and matching the total absolute forces provided from the test data. For the same wind shear profile and various wind headings, surge and sway loads between wind tunnel model tests and numerical simulation were in excellent agreement. However, the time domain model did not reproduce the expected heave load and yaw moments; however, these loads were not significant compared to hydrostatic stiffness and cable reaction loads, which prevent any significant motion or deflection. Figure 11: Ferry and pontoon model for time domain docking simulations Environmental loads on the pontoon were computed with the use of a ShipMo3D hydrodynamic model which includes wave radiation and diffraction effects. The resulting loads include hydrostatic buoyancy, wave excitation, current, and wind loads. In order to model the effects of impacts during docking procedures, a directional elastic force constraint was placed between the contact point of the pontoon and the ferry to represent the fender. The contact force is only applied as the ferry and pontoon interfere in a contact event. The contact force uses a constant stiffness. Limited damping was used to help ensure a conservative result in computing the impact load on the pontoon. The total yaw moment and reaction forces were computed at the geometric centroid of the dock. VERIFICATION WITH PHYSICAL DATA The wind and hull resistance coefficients were validated with physical model test data. The dynamic time domain ferry response in waves was also compared with model scale tests in a wave basin facility to ensure validity. Data of wind loads produced from wind tunnel tests on a scale model abovewaterline superstructure of the ferry was provided by Oceanic Consulting Corporation (Oceanic) and BMT Fluid Mechanics, as shown in Figure 12. The corresponding superstructure surge and sway wind coefficients to be used in the time domain simulation models were established by comparison of absolute wind load with the wind tunnel test data. The variation in wind Figure 12: Wind tunnel testing completed by BMT Fluid Mechanics In the model scale tests, a planar motion mechanism (PMM) was used to drive the scale hull model through the water at a range of forward speeds and various static hull yaw angles, as shown in Error! Reference source not found.. The hull resistance tests indicated skin friction forces on the hull that resisted forward speed while the drag tests gave information of lateral drag loads that would occur from relative currents. In general, surge viscous loads are important as they will influence the power required to drive the ferry forward and sway viscous loads are important as they influence the lateral deflection of the system. The yaw angle was varied slightly to give an indication of the load variation that may occur in the event that the guide and traction cables constrain the ferry to some offset yaw angle as the ferry moves through the crossing. A comparison was made in steady state conditions in the time domain model. In general, the loads are in agreement for small yaw angles. However, due to the simplicity of the hull resistance time domain model, no yaw moment was produced as the yaw angle increased. In addition, the forward surge resistance decreased as the yaw angle increased in simulation, while the model test data indicated an increase in surge resistance over the range tested. 300

8 Surge drag loads must eventually drop to zero once the yaw angle reaches 90 degrees and so the model scale test results highlight the more complex flow effects at moderate yaw angles. However, time domain crossing simulations indicated the dynamic yaw response of the ferry remains within 5 degrees of the forward direction and so the difference in surge drag is not significant. Furthermore, the lack of yaw drag moment was not a significant concern due to the significant yaw stiffness imparted by the cable system. position and orientation relative to a wavemaker in a wave basin facility as shown in Figure 15. A top view of the beam wave loading case can be seen in Figure 16, which shows a combination of the model scale test and a snapshot from the time domain simulation model. In the time domain model, the mooring lines used had the same material properties as the full scale cable system. However, to ensure physically consistent results, additional linear springs were added at the anchor points of the cables as well as a pretension value that corresponded to the model test system. The time domain model used a JONSWAP spectrum as in the model test case. Figure 15: Model basin testing of cable ferry hull Figure 13: Steady state simulation (line) and wind tunnel test loads (dots) as a function of relative wind direction to the ferry The model scale tests provided data corresponding to lateral current loads of 1m/s. A comparison was made in steady state conditions in the time domain model. The surge and sway lateral drag loads compare reasonably well over a wide range of yaw angles. The yaw moment was not accurately represented in the simulation model though it is not significant due to the substantial yaw stiffness of the cable system used by the ferry An initial comparison of the wave basin tests and the time domain model showed the roll response of the time domain model in beam waves was less than the seakeeping model tests. Sensitivity studies indicated that the roll wave radiation parameters used in the time domain simulation, which produce a damping effect, were too large. The wave radiation data was scaled to better match the model test results and ensure realistic roll behavior. All of the tests were performed at a significant wave height of 1.2m and peak period of 5s. The focus was on comparison of standard deviation of response between the model tests and time domain simulation results. Accelerations at several locations on the deck and motion of the center of gravity were very similar. Figure 14: Resistance and current drag testing using PMM To assess the seakeeping response of the ferry, model tests of the ferry in waves were carried out by Oceanic. The experiment model scale setup consisted of a soft mooring to maintain ferry Figure 16: Seakeeping setup: wave basin model schematic shown on left and ProteusDS model setup shown on right. 301

9 DEVELOPMENT OF DESIGN BASIS Knowledge of the environmental conditions at the site is needed for the analysis. Wind, wave and current conditions will affect the hydrodynamic loading on the cables as well as the ferry and the terminal. An analysis of environmental conditions was completed by Cascadia Coast Research Ltd. The 1, 50 and 100 year wind, current and wave conditions were determined for the site. Expected 1, 50 and 100 year extreme wind speeds were estimated for 8 directional ranges between 0 and 360 degrees. Values were produced for the terminal location. The USACE Coast Engineering Manual suggests that factor of 1.2 can be applied to determine the wind speeds at the mid-point of the crossing. Extreme current values in the northerly and southerly directions were determined using data from a 3 month deployment of a buoy-mounted ADCP (acoustic doppler current profiler). In this particular region, the tide levels are tidal driven and deterministic. The surface currents are driven by wind and are probabilistic. For this project, current extremes were estimated statistically based on 3 months of surface measurements using a peak over threshold approach. The relatively short data set resulted in a fairly large predicted current relative to the maximum measured value. Circulation modeling approaches might be used to more accurately predict the 1, 50 and 100 year extreme current values. Wave measurements were made at the site using a wave buoy using two deployments. The SWAN (Simulating Waves Nearshore) wave model of Baynes Sound was developed. The model used wind data from a nearby weather station as input. The model was validated by comparing the mid-channel buoy measurements to wind inputs. The model was used to hind-cast wave conditions at each of the terminals and at the crossing midpoint. The waves are strongest at directions of 90 and 135 degrees due to the fetch and winds in these directions. The 1 year return period conditions were used to analyze the possibility of operating in a once-per-year condition. For example, vessel accelerations and roll must be acceptable under these conditions. The 50 year and 100 year extreme conditions were used to assess the system design. This included a damaged conditions scenario, where one guide cable has been released, and the other guide cable and traction cable must take the load. When using dynamic analysis, safety factors for the damaged condition and the intact condition are specified in API RP 2SK as 1.25 and 1.67, respectively. These values were used to justify the use of a safety factor for the mooring design of 1.3 and 2.0, respectively (API, 2008). A safety factor of 5.0 is also applied to the pretension value used in the cables. Figure 17: RAO response at 4m/s forward speed and 90 deg absolute wave direction; note the yaw natural period due to the extra cable stiffness in the yaw response. Table 4: Wind, wave and current conditions considered in analysis 1 year 100 year Extreme Wind speed (m/s) Significant wave height (m) Peak period (s) Current speed (m/s) FREQUENCY DOMAIN ANALYSIS The numerical wave radiation and diffraction analysis was completed with ShipMo3D. A database of hydrodynamic coefficients was produced that enable wave radiation and diffraction effects to be incorporated into time domain simulation through the use of the Cummins equation (Fossen, 2011). The wave radiation and diffraction loads are resolved 302

10 through frequency domain analysis and are therefore inherently linear. While the draft of the ferry was relatively shallow, the large waterplane area and barge hull form is well-suited for potential flow analysis as wave radiation tends to dominate hull viscous effects in damping vessel motion. The nonlinear time domain model was used to numerically compute a linear constant spring stiffness to represent the cable reaction loads at several locations along the crossing to ensure interaction with the bottom was considered. No linear equivalent damping from the cables was incorporated and this was deemed conservative since it would reduce the ferry motion in the frequency domain results. In addition to generating the hydrodynamic database, the response amplitude operators (RAO) that indicate the ferry motion as a function of incident frequency and ferry forward speed were computed. The resulting RAOs were used to assess ferry motion for various weather directions and forward speeds. The most important result indicated that at maximum forward speed, opposing seas from waves in the 90 degree absolute heading (see Figure 3) could generate unacceptable pitch and roll motion. This is indicated by RAOs in Figure 17. However, this effect is significantly reduced as the ferry forward speed decreases and can therefore be addressed by operational restrictions. Another important result was the cable stiffness indicated a yaw natural period of approximately 15 seconds, as shown in Figure 17. However, this is too far above the expected ocean wave periods to induce any motion. There were relatively minor variations in RAO values along the span of the crossing, which indicates the interaction with the sea floor plays less of a role than pretension in influencing ferry response in the linear range. The RAOs produced were also useful as a further verification of the time domain analysis completed. (NPD) wind spectrum model was used. The wind model models wind gusts based about the 1 hour mean wind speed. For each simulation case, three wave seeds were used to generate three unique random wave realizations. Each wave seed is used to generate a randomized wave surface based on input parameters of significant wave height, mean wave direction, spreading function and wave period. This was done to ensure that statistically consistent results are being realized and that accurate maxima are determined. Transit Analysis The nonlinear time domain simulation analysis of the ferry transit completed produced data that was used to guide the ferry structural design, selection of the cables, and design of the ferry terminal. The transit simulations were conducted using the 50 and 100 year return period conditions with coincident extreme wave, wind and current conditions from 90 and 135 degrees. Loading from other directions would be significantly less due to the reduced wave heights. Simulations of the ferry crossing both to and from Buckley Bay were completed. The loading varies depending on the heading of the vessel (following vs. head seas). To complete the simulation at the target speed it was necessary to simulate the winch behavior. The ferry is propelled by increasing the tension in the traction cable. To apply this tension a PID controller was used. The PID controller monitored the ferry forward speed and adjusted the force applied to the cable to mimic the bull wheel system. This controller allowed a desired transit speed to be set for the simulations. Simulations were completed at 3.9 m/s and 1.9 m/s. TIME DOMAIN ANALYSES Time domain analysis of the ferry while in transit, while stopped during the crossing, and while docking was completed. The transit analysis was used to assess operability of the ferry in the extreme conditions and determined input values for strength analysis. The analysis of the ferry response while stopped during the crossing was used to assess braking system loads. The docking analysis was used to assess the ability of the ferry to offload vehicles in extreme conditions (by examining relative roll) and provide data for detailed design of the terminal facilities. Wind, Current And Wave Models For each simulation, a JONSWAP wave spectrum was used with 1200 wave segments to ensure a non-repeating random sea condition. A constant uniform current was applied through the water column 1. As recommended by several offshore standards (API, 2008; DNV 2010), the Norwegian Petroleum Directorate 1 A log or power law current profile that tends to 0 m/s near the seabed should be used in future projects to better model the current profile. (Soulsby, 1997) Figure 18: Dynamic tension shows very large period (~600 sec) from ferry lateral offset and cable uplift during transit, medium period (~50 sec) due to wind load on the ferry in sway, and high frequency (~2sec) due to wave load on the ferry The effect of cable pretension on the cable ferry was also checked with the time domain simulations. Figure 21 (at end of paper) shows the typical drift from the crossing centerline. A maximum lateral drift of 92m was found with a pretension of approximately 200kN. The amount of pretension in the system 303

11 affects the excursion distance in extreme conditions from the transit centerline. It also affects the maximum tensions observed, and the ferry roll motion. The maximum tension in the cables during the 100 year storm was found to be 571kN in the beam sea loading case. Several simulations were executed to observe the impact of increasing the pretension in the system. A catenary shape provides some compliance to the cable ferry system, allowing it to respond to changing conditions without severe changes in cable loads (e.g. snap loads), as would be the case in a taut mooring system. However, too much tension results in high cable loads in extreme conditions and a stiff system response. The higher the pretension the more effect the cables have on the vessel dynamics. The dynamic tension in the cables from wind and wave loading during a portion of ferry transit can be seen in Figure 18. The ferry roll motion was determined during the transit simulations. Maximum roll angles of between 8 and 12 degrees were observed for the ferry transits. Interestingly, at higher wind speeds, the larger mean cable tensions reduced roll motion of the ferry, resulting in lower roll angles than some simulations with lower wind speeds. Table 5 presents a summary of the transit simulation results. The key data produced through the transit simulations were: i. Bull wheel / traction winch system power and load limits ii. Loads on fairleads on the ferry iii. Ferry roll motion iv. Lateral deflection during the crossing v. Ferry accelerations at aft/forward port and starboard corners and in the passenger lounge vi. Air gap during crossing vii. Maximum cable reaction loads at anchor points viii. Maximum cable tensions ix. Maximum cable tensions and ferry roll motion were assessed in a damaged state (one guide cable released) Table 5: Summary of transit simulation results Max Condition Cable Drift Load (m) Pitch ( ) Roll ( ) (kn) 100 Year Storm Extreme Storm Damage Case Vertical Acceleration (m/s 2 ) Braking Analysis The ferry s capacity to perform an emergency stop in the 100 year wave and current conditions was simulated. The maximum stopping distance in this emergency situation was assessed given a maximum traction winch force available for braking. The ferry was able to stop within 65 m (less than one ferry length) during 100 year wave and current conditions and with conservatively high 55 knot winds, given a traction braking load of 69 kn. Emergency stop simulations were completed with the ferry at locations corresponding to 20%, 40%, 60% and 80% of the crossing distance. The brake was applied on the traction cable and the maximum braking loads observed. This corresponds to a friction brake that might be applied to the traction winch system. The maximum loads in the cable were measured, and a maximum of 420 kn was found. The stopping distance did not significantly vary, which in these loading conditions indicates the inertia effect of the cables did not have a significant impact compared to the inertia of the ferry. Damaged Condition Analysis The cable ferry system was tested to determine the response when one of the guide cables has failed or parted due to some unforeseen event. For the analysis a guide cable was removed and the brakes applied to the cables. The tensions in the cables observed when the ferry was at a position corresponding to 40% and 60% of the crossing in the 100 year storm conditions. The tensions in the cables were observed to ensure that at a safety factor of greater than 1.3 was achieved to avoid the remaining cables from breaking. The maximum cable tension measured in the damaged case was 691 kn. Docking Analysis The docking analysis consisted of two primary sets of test cases. The first set analyzed the effect of the ferry impacting the pontoon, and the second analyzed the interaction of the ferry and dock while the ferry was docked with the pontoon. The setup of the cables, pontoon and ferry is shown in Figure 20. The piles which restrain the pontoon from moving in the simulation are not shown. In the impact study, several simulations were run with the ferry impacting the dock at 2 knots. The simulations started with the ferry 100m from the dock. The PID winch controller was placed under speed control to drive the ferry forward. These tests revealed the impact loads given a bumper stiffness. In these simulations the loads on the dolphins (pile groups) are assessed. This approach allowed for careful assessment of the complex impact and ferry sway loads. In addition, the angle of approach of the ferry is assessed. These tests were designed with the purpose of setting the structural load requirements for construction. The maximum surge force on the pontoons during impact were ~3300 kn and the maximum sway load on the pontoon was 2800 kn. The pontoons must withstand a total yaw impact moment of knm. These forces are the maximum instantaneous loads during impact, and the shore structures do not need to withstand these loads for long periods. The second set of simulations conducted for the docking analysis focused on the dynamics of the ferry and dock during extreme conditions. The relative roll of the ferry and the dock 304

12 were assessed to ensure emergency offloading in the 100 year storm event. A wing wall is often used to ensure alignment of the ferry during berthing and to reduce relative roll of the ferry. Simulations were additionally used to assess the effect of wing walls on the sway loads and yaw moments applied to the dock. It was found that the wing wall reduced the relative motion of the ferry by approximately 50%. The heading of the ferry relative to the dock was also measured in the simulations. This was done to ensure that the off-loading ramp on the ferry had sufficient overlap even when the ferry was being impacted by gusting wind and irregular waves. The maximum relative heading between the pontoon was found to be around 8-12 in the 100 year storm and extreme conditions, and 2-3 degrees in the 1 year storm. The relative roll during offloading conditions (1 year storm) was found to be a maximum of around 4 degrees. FATIGUE ANALYSIS An investigation was completed to assess the influence of environmental loading on the fatigue life of the cables. This consisted of checking the fatigue life of the cables due to the dynamic tension that resulted from storm condition wave, current, and wind loading on the ferry as it crossed the channel as well as separately investigating the effects of current load and the resulting Vortex Induced Vibration (VIV) damage. Rather than accurately computing the fatigue life of the cables, a simpler but more conservative approach was used to compute the lower bound of resulting fatigue life to assess if there was any risk of short-term failure due to fatigue. The process in mooring standard API RP 2SK was used to compute the fatigue life in storm conditions as the ferry crossed the channel. The process consists of using the rainflow time domain cycle counting method to establish the expected value of tension range during the crossing. The total fatigue life was computed assuming the ferry is always crossing, the 100 year storm acts 50% of the time, and the extreme storm condition acts the remaining 50% of the time. In spite of these very conservative assumptions, the resulting fatigue life was resolved at 89 years for the wire rope selected. To compute the fatigue life due to vortex shedding on the cables and the resulting VIV that could occur, it was assumed that the cables had uniform and constant tension. It was assumed the maximum wind-induced current velocity of 1m/s (vs a tidal maximum of only 0.2m/s) was constant throughout the water column and that it acted 100% of the time. This is conservative because higher currents will induce faster cable vibration and larger bending stresses. Even with these conservative assumptions, the range of bending stress induced was too small to have any significant impact. The resulting fatigue life due to VIV alone was many orders of magnitude longer than the expected fatigue life due to wind and wave loading. The fatigue analysis did not consider the abrasive damage due to interaction of the cables with the sea bottom, winch equipment, dock infrastructure, or general handling. These effects are expected to be important but the cables can be regularly inspected for excessive wear due to these effects. FERRY SERVICE Using the data from the analysis, the service expectations of the ferry were estimated. The service expectations were determined based on an analysis of the frequency of occurrence of certain wave sizes, the roll and motion response of the ferry, and the strength of the securing cables. For the Denman Island Ferry, it was determined the ferry would be able to maintain a highly reliable service (99.4%) and that fewer than 35 trips a year out of approximately 12,400 trips would be affected to the extent that a trip might be delayed due to conditions. Two operating modes were assessed, docked and transit. For each of these modes an operation limit was assessed. The limits presented in Table 6 were used to determine at which point operations must cease based on the weather conditions. Table 6: Motion and seakeeping criteria Vertical Acceleration Roll Criteria <0.275g RMS at any point on the ship +/- 4 degrees RMS +/- 2.5 Degrees RMS Pitch +/- 1.5 degrees RMS Deck Wetness CONCLUSIONS < 5% at the top of the bulwarks <0.10g RMS at any point within the passenger accommodati on +/- 8 degrees Significant +/- 5 degrees Significant +/- 3 degrees significant Condition Restricted Operability Restricted Operability Caution Range Restricted Operability Restricted Operability A comprehensive dynamic analysis was completed on a 2000m cable ferry. The analysis consisted of the following: Model basin resistance tests Wave basin roll motion tests Wind tunnel topside drag coefficient tests Frequency domain seakeeping analysis Environmental conditions determination Time domain model verification Time domain transit simulations Time domain docking simulations Time domain braking simulations Time domain damaged condition simulations Cable fatigue analysis 305

13 Each of these items provided key data for the detailed design of the ferry and the terminal. These were the key objectives of this study: Determine the effect of ferry motions on the stresses on the cable and resulting stresses in the anchoring system Determine the motion response of the ferry due wind and waves Evaluate the environmental conditions at the site Determine the ferry interaction with the pontoon and determine the loads and relative motions of the ferry and the pontoon Determine the powering requirements and drive system for the ferry Determine ferry service levels The methodology followed provided extensive data which was critical to determining the level of service of the ferry and ensuring safe and reliable system design. The numerical modeling approaches taken were found to be very effective in assessing the ferry dynamics. ACKNOWLEDGEMENTS The authors wish to acknowledge British Columbia Ferry Services, Ltd for permission to publish the work completed. REFERENCES API RP 2SK, American Petroleum Institute Offshore Standard, Design and Analysis of Stationkeeping for Floating Structures, DNV OS E301, Det Norske Veritas Offshore Standard, Position Mooring, October Fossen, T. I., Handbook of Marine Craft Hydrodynamics and Motion Control, Wiley, McTaggart, K., Roy, A.,, D., Nicoll, R., Perrault, D., Simulation of Small Boat Launch and Recovery from a Ship with a Crane, ASNE Launch and Recovery Symposium 2012, November 14-15, 2012, Maritime Institute of Technology and Graduate Studies (MITAGS), Linthicum, MD. McTaggart, K. A., Verification and validation of ShipMo3D ship motion predictions in the time and frequency domains. In International Towing Tank Conference Workshop on Seakeeping: Verification and Validation for Non-linear Seakeeping Analysis, Seoul, Korea, Nicoll, R. S.,, D. M., Attia, J., Roy, A., Buckham, B. J., Simulation of a high-energy finfish aquaculture site using a finite element net model, Proceedings of the 30th International Conference on Offshore Mechanics and Arctic Engineering, Rotterdam, The Netherlands, June 19-24, Nicoll, R. S., Wood, C. F., Roy, A. R., Comparison of physical model tests with a time domain simulation model of a wave energy converter, Proceedings of the ASME th International Conference on Ocean, Offshore and Arctic Engineering, OMAE 2012, July 1-6, 2012, Rio de Janeiro, Brazil. Soulsby, R., Telford, T., Dynamics of Marine Sands, 1997., D., Buckham, B.J., Nicoll, R., Design Through Simulations: Finite Element Capabilities for Ocean Engineering, Proceedings of the 27th International Conference on Offshore Mechanics and Arctic Engineering, Estoril, Portugal, June 15-20, Figure 19: Figure showing cables interaction with seabed 306

14 Figure 20: Docking-model setup Figure 21: Typical drift drawing showing the deflection of the ferry from the crossing centerline. A maximum deflection of 92m was determined. 307

15 Discussion John Boylston, Fellow When we read this paper we were very impressed as to the level of analysis, model testing and engineering that had gone into what is not a new technology, but a significant increase in scale. The authors are to be congratulated on taking little available base data and working it into a logical approach for determining base design parameters. Researching the proposed ferry system, we found there to be some adverse public reaction, and therefore BC presentations and public testimony from where additional information could be obtained. While we have no problems with the numerical analysis and using whatever standards are available, the fact is that there is evidently no full scale operational data for the proposed system, or anything like it. Scaling up smaller cable ferries, operating on single cables, would not be realistic. As complete as this effort is, we submit it still seems like a risky basis upon which to invest a considerable sum. There are some specific design aspects that should trouble BC Ferries for such an undertaking where the invested cost of the new cable ferry system will be similar to the cost of a replacement ferry, similar to the existing Quinitsa. Some of these concerns are as follows: - Cable Strength and arrangement - It is indicated that the existing factor of safety for the cable system (3 cables) is 2.0 and if one guide cable lets go, the safety factor for the two remaining cables drops to We suggest this is insufficient. - The three cable system - The system seems to have attributes; however, there should also be some concerns. As the guide cables are not constrained on the ferry, but the hauling cable is always under tension, pulling the ferry along, it would seem assuming that all three cables equally support the ferry is unrealistic. It would appear that the hauling cable would bear the brunt of the sharp impact loads one could expect from seas upon a barge form. - Operationally, stability should be more fully investigated. In the US, we had problems with early offshore supply boats of a barge form, where the low draft and freeboard would allow the deck edge to submerge in modest beam seas / winds and allowed lifting of the windward chine out of the water. In many cases, this resulted in flipping the vessel upside down. We adopted the more stringent Rahola stability criteria which should be applied here. For this particular case, if a windward guide cable failed in a beam sea / wind, the added tons on the leeward side from the remaining guide cable, would give an initial list, we estimate, of about 3 degrees and only 4 degrees is needed to submerge the deck edge. The restraint of the leeward guide cable adds a force to further prompt over turning. With such a low freeboard and wet decks the consideration that cars might start sliding to the leeward side, compounds the problem. We have some suggestions to address some of these concerns: 1. Install 2 large guide cables (perhaps 2.5 inches) on the present run and fit the Quinitsa with guides on each side. The ferry would still be powered, but would allow measurement of operating data. This data collection and operation would increase public confidence. The cables could then be used for the new cable ferry. 2. Consider two heavier guide cables as above with a much lighter, more flexible, (cored?) traction winch cable. The existing proposed cable size around a 2 sheave traction winch with the minimum diameter suggested will make a short life of the type of cable specified. 3. Rent a powered traction winch and install it, temporarily, on the Quinitsa, running its cable ashore to a fixed point. Then a determination can be made of the force necessary to move the Quinitsa at different speeds. We do not believe the savings envisioned from barge construction will be realized after reading the Technical Statement of Requirements. A ferry with barge form is specified, not a barge. Lapped plates will not save that much and will lead to premature hull corrosion at the laps. We believe that a straight barge form is not well suited to this application. If a modified, more ferry like form, hull had a spoon type bows and the receiving pontoons had identical spoon indentations, it would be possible to slightly ground the ferry at each terminal, taking away relative movement. A well presented paper. Vince den Hertog, Member The authors are complimented on the clearly written and interesting paper, as well as the thorough approach taken in numerical modeling and validation with model testing. Developing a numerical time-domain simulation methodology to analyze the dynamic cable behavior as part of the cable ferry design process is welcome, and indeed necessary: Accurately predicting the variation in tension in the haul cable as it is influenced by pre-tension, accelerations, decelerations, transit speed, environmental forces, and catenary bottom bathymetry is difficult or imprecise by any other method, yet all these aspects should be considered as part of design process for any cable ferry, especially on long crossings that are subject to variable operating conditions. In this discussion, I would like to take the opportunity to expand upon how the approach outlined in the paper has the potential to significantly help the ferry designer evaluate the implications from pre-tensioning of the haul cable (the cable that passes 308

16 through the haul winch). Haul cable pre-tensioning can have a strong effect in a number of areas but, in particular, on the mechanical strength and longevity of the winch itself. operating profile are needed to make sure that the winch is not only strong enough but has the necessary bearing and fatigue life. As explained in the paper, it is often the case that the haul cable must be pre-tensioned in some way to maintain a minimum tension at the tail-end (lower tension end) of the haul winch at all times so that the cable does not slip over the bull wheels, especially during ferry accelerations and decelerations. The weight of the cable alone from catenary is not always enough. The minimum tail-end tension necessary depends on the winch design: In general, the more cable wraps the winch is designed to hold (i.e. more contact area), the lower the minimum tail-end tension needs to be. Pre-tensioning is typically maintained with a pulley and weight arrangement fitted to the ends of the haul cable on shore, or is set in one shot by pulling in & clamping the haul cable at the shore termination at the desired pre-tension. However, over long crossings where a significant amount of cable lies on the bottom, it is virtually impossible to maintain constant pre-tension levels over the entire run. When the ferry is stationary (and not constrained by the dock), the tension in the haul cable starts off being more or less the the same on either side of the winch due to pre-tensioning and catenary. However, as the ferry begins to move, the tension in the haul cable on the side that is being hauled in (the lead-end) becomes higher than the tension in cable being paid-out on the other side of the winch (tail-end), with the difference being roughly equal to the propulsive force. From then on, as the ferry makes its way over the crossing, it is important to realize that even when the propulsive force remains the same (say, with a constant speed), the average or prevailing tension in cable on both sides of the winch can vary considerably with, the amount of cable lying on the bottom, catenary, environmental considerations and the other factors discussed by the authors. Determining what the appropriate level of shore-end pretensioning should be applied to maintain minimum levels to prevent slippage in all situations can become a very challenging problem indeed, given all the variations that can occur over the crossing. That said, arbitrarily applying too much pretensioning in the system can lead to problems for the winch, since it directly influences the average tension over the crossing, and this in turn affect forces acting on the winch. Cable ferries typically use winch designs with either a single bull wheel or dual bull wheels. On single bull wheel designs, the cable is wrapped multiple times to achieve sufficient contact area. On Dual bull wheel designs, as shown Figure 4 (repeated below), the cable is lead back and forth from one bull wheel to the other multiple times. The design of dual bull wheel winches is particularly sensitive to pre-tensioning levels since the arrangement is like a block and tackle: tensions over the bull wheels have a multiplying effect in the total force pulling the bull wheels together. Consequently, subtle changes in tension can significantly affect the forces on drive shafts and bearings. Because of this sensitivity, accurate predictions of the haul cable tensions on both sides of the winch throughout the ferry To summarize, the sophisticated time domain simulation methodology outlined in the paper seems to provide the means to predict the dynamic variations in both lead end and tail end cable tensions at the winch over the crossing under all environmental conditions. This is invaluable not only to ensure that slippage does not occur during starting and stopping, but also to verify that the haul winch is strong enough and has the longevity needed. To validate the methodology further, if this has not been done so far, the authors are urged to make use of any opportunities to make comparisons of predicted tensions at the winch to actual measurements made on other cable ferries in service, perhaps through the use of clamp-on cable load cell tensiometers installed on either side of the winch. Such measurements have been made successfully in the past on other cable ferries. If there is an opportunity to do the same on the Denman Island ferry once in service, it would certainly compliment the excellent work done by the authors so far. Dan McGreer, Member, and Chad Oldfield, Member The authors are to be congratulated for preparing a very interesting and useful paper about a dynamic simulation of a cable ferry. The analysis is very comprehensive and gives a lot of insight into cable ferry dynamics. We have a few questions for the authors. 1. From the figures in the paper the wind loads appear to be calculated based on the ferry without vehicles on the car deck. Have the authors also carried-out wind load calculations with vehicles included? 2. The seakeeping model tests showed that the roll motions were larger than predicted by the time domain simulation. The authors indicated that sensitivity studies indicated that the roll wave radiation parameters resulting from the panel 309

17 size were the reason. Was there a comparison of the viscous damping between the numerical model and the tank test model and could this also be a possible reason for the difference? 3. In the section on frequency domain analysis the authors indicate that the yaw natural period is approximately 15 seconds. The cable tension will affect the yaw natural frequency as increases in tension will reduce the period. In what loading case is the natural frequency 15 seconds and how does the yaw period change with tension? 4. In the docking time domain simulation the dolphins supporting the dock are fixed. In reality the dolphins will are flexible and will possibly reduce the impact forces on the pontoon. Have the authors investigated the possible effects of dolphin stiffness on the results of the docking simulation? 5. Are the simulation results sensitive to the elasticity of the cable? How accurate is the cable elasticity and can it vary with time? 6. Finally, does the fatigue analysis include the stresses caused by the bending stresses as the cable goes around the bullwheels? Alan Winkley, Life Member Cable ferries are a hybrid type of vessel incorporating some aspects of basically stationary vessels, such as moored floating offshore oil rigs, together with what otherwise would be a conventional free-floating ferryboat. This paper illustrates how the design of cable ferries has advanced through application of quite sophisticated hydrodynamic analysis. The authors describe a system comprised of vessel, cables, docks, shoreside cable attachments, the seabed, and, of course, a spectrum of wind and waves that has some probabilistic relationship to the specific ferry crossing. Evidently, this sophisticated approach to the problem of cable ferry hydrodynamics has it roots in dynamic analysis of offshore oil rigs and other floating structures that are moored. The traction winch technology probably stems from the offshore oil and gas industry, as well. It is interesting to note how the hightech analytical methods of the offshore industry have filtered down to what might easily be perceived as a mundane, even antiquated, means of short-haul transport. I have a few questions: 1. Are the authors aware of this type of dynamic analysis (either time-domain or frequency-domain) being carried out for any other cable ferry anywhere in the world? 2. Would the proposed Denman Island ferry be the biggest cable ferry ever, in terms of vehicle capacity, displacement, crossing distance, or any other parameter? 3. Was thought given to locating the traction winch and associated cable works to either the port or starboard side, rather than along the centerline of the hull? 4. Addressing the real-world cable-ferry operation (as opposed to the computer model simulation), how can the tension in the traction cable be deliberately increased or decreased? 5. Referring to the second sequential diagram in Figure 19, it appears that a large portion of each cable leading aft from the ferryboat forms a rather shallow catenary which doesn t touch the seabed. Does this slightly submerged cable present a hazard to vessel traffic moving perpendicular to the ferry crossing? 6. I am confused by the presentation of motion and seakeeping criteria in Table 6. Does one column under the Criteria header apply to the docked condition and the other column apply to the transit condition? The following comments pertain to the analytical model used for the analyses: Under concept validation (p. 4) it says: The approximate weights of the cables were included in the lightship weight. It seems to me that, at no point during a crossing, would more than one-half of the weight of each cable be borne by the ferryboat. The remainder would be supported by the shore attachment, the floating dock, and the seabed. In fact, while traversing shallow water and with low cable tension, a large fraction of the cable weight would be borne by the seabed. Furthermore, the cables are mostly submerged, so that their weight in water is less than the weight in air. For the stated unit weight of cables (7.0 kg/m), this would lead to over-estimating the displacement of the ferryboat by more than 21,000 kg. Judging from Figure 9, the wind force estimation does not take into account the sail area of vehicles parked on the deck of the ferryboat. This would tend to under-estimate the potential heeling moment and also the tension in the guide cables. There is no doubt that a traction-cable drive is a more efficient means of locomotion than any type of hydrodynamic propulsive device. This would be a point in favor of using this form of drive for any cable ferry. Nevertheless, in my experience, I have noted a reluctance on the part of winch builders to get involved with cable ferry projects. This probably stems from the winch builders aversion to doing research and custom engineering for a one-of-a-kind cable ferry application. They are indeed rare applications. And, judging from my discussion with operators of cable ferries, it takes a great deal of postdelivery adjustment and modification to achieve reliable operation. That said, has the design progressed to the point of getting one or more winch builders closely involved? I believe that is critical. I complement the authors on a big addition to the technical literature on the subject of cable ferries. 310

18 ADDITIONAL REFERENCE WINKLEY, A Design of a Cable-Ferry for Oregon s Willamette River, SNAME Small Craft Symposium, Ypsilanti, MI, USA. William A. Wood, Member Reader: Andrew Lachtman, Associate Member This interesting paper addresses a very small and largely ignored segment of the ferry industry - cable ferries. Like all natural ferry operations, each individual operation represents a unique solution to the age old dilemma: How do I get to the other side? In some cases a cable ferry, rather than a freely maneuvering ferry, is the solution that makes the most sense. As the authors point out, cable ferries are not complex, have low initial construction and maintenance costs, are fuel efficient, have minimal crew requirements, have simplified navigation and berthing requirements, and are low risk. Cable ferries also have some disadvantages that must be considered: (1) a cable ferry has only limited maneuverability in order to avoid a collision; (2) the cables rising to the surface can interfere with the navigation of other vessels; (3) debris such as trees can hang up on the cables interrupting service; and (4) the ferry must be disconnected from the cables and towed to a shipyard for maintenance and repair (during which time a replacement ferry is most likely not available to continue the service). Cable ferries tend to be either cable guided or cable powered. Cable guided is when the ferry is attached to a guide cable, but is either self-propelled or propelled by a separate yawl boat or tug or by a river current. Cable powered is when the ferry uses a single cable to both guide and pull itself across the waterway. The cable ferry addressed in this paper is a combination of both, having two guide cables and an independent powering cable. The proposed ferry between Buckley Bay and Denman Island is large and has a long crossing compared to more traditional cable ferries. It would be interesting if the authors could comment on why a cable ferry is being considered for this route. The existing ferry appears to be a freely navigating vessel. The existing terminals appear to have floating bridge ramps oriented so the ferry docks into the 135 o prevailing wind, wave and current direction. Will the existing terminal facilities be replaced, and will they be reoriented as shown in Figure 21 to be more perpendicular to the 135 o direction placing the docked ferry at roughly 90 o to the prevailing wind, wave and current directions? The studies undertaken by BC Ferries and the authors were obviously considered necessary to answer this and many other questions, and to prove that a cable ferry will work for this route. The answers could give assurance that the advantages of cable ferries (low cost and low risk) can be achieved. In the absence of a design methodology for cable ferry dynamic analysis, the authors have done a fine job of creating one. They used hydrodynamic and wind tunnel model testing, nonlinear finite element cable modeling, and several methods of dynamic analysis to simulate the ferry crossings and docking under a range of normal and extreme environmental conditions. Through these studies they have shown, with reasonable certainty, that the proposed cable ferry can work for the Buckley Bay Denman Island route. It would be interesting to know if the same results could have been achieved using quasi static methods. Is BC Ferries also planning to replace the freely navigating ferry between Denman Island and Hornby Island with a cable ferry? If BC Ferries does indeed install a cable ferry on this route, it is hoped that a program of full scale testing and correlation of the ferry motions and cable dynamics will be undertaken. Those results, when compared to the analyses, should be made available to the SNAME community in a future follow up paper. I would like to thank the authors for a very interesting paper on a rarely addressed subject. Authors Closure The authors wish to express their appreciation for the thoughtful review by the discussers and their interesting questions. The positive comments are encouraging and we appreciate the feedback of additional suggestions. We also thank SNAME for publishing this paper. John Boylston asked Are all three cables sharing the load. The cables share the load, but unequally depending on the motion of the ferry. The time domain simulation takes this into account, for example if the vessel was rolling one cable would see a slight reduction in the tension while the opposite cable would see an increase. Since the cables are pre-tensioned there is never a situation where one cable is completely unloaded and therefore not able to contribute to increased tensile forces, Mr. Boylston commented that the safety factors are low. It is true that the factors of safety appear low, however with the higher level of certainty of the dynamic analysis, it is accepted practice to lower the factors of safety. The factors chosen exceed international guidelines (such as API RP S2K) utilized for offshore moored platforms. It is important to realize that the conditions which result in the lowest safety factors are in fact conditions where the ferry would not be operational or in a very limited service. The scenarios replicating a broken cable are imposing conditions on the ferry worse than have ever been seen in service and this has the effect of increasing the factor of safety Mr. Boylston discussed a number of stability concerns regarding the ferry and suggested that the Rahola criteria be adopted. We have done a comparison of the ferries stability characteristics 311

19 compared to the Rahola criteria and they are summarized in the table below. The stability criteria for a passenger ferry in Canada is more rigorous than the Rahola criteria in that we have larger area requirements and also wind/passenger heeling superimposed on top of damage cases. BC Ferries also requests two compartment damage survivability which is in excess of the rule requirement. Mr. Boylston suggests that a cable break on the leeward side would impose 30 to 40 tons of additional load and cause 3 degrees of heel. In fact all three cables weigh a total of 44 tonnes and only ½ of the total weight is carried by the ferry in any loading case. The heel of one cable lost in conjunction with a 50 tonne truck to one side of the deck is only 1.5 degrees. We are satisfied that the stability is adequate especially considering the benign environment that it will operate in. Note that Rahola criteria is only for intact stability, these criteria were met with one compartment flooded in conjunction with wind heeling and passenger crowding Rahola Criteria Attained Pass/Fail Righting Arm 2.625m Pass Greater degrees Righting Arm 2.263m Pass Greater than 30 degrees Area under 0.50m-rad Pass Righting Lever Curve> 0.08m-rad to Downflooding Mr. Boylston suggested a test program using an existing ferry to get real life data. BC Ferries evaluated this prior to commissioning the study and determined that the program would not be cost effective. There was a comment that barge type construction would not be cost effective and would result in early corrosion occurring in the lapped joints. Unlike most barges, these ferries void spaces are completely dry and fully painted and ventilated inside. Furthermore, the welds are completely sealed by welding all round. We have many ferries in the Maritimes with this type of construction approaching 30 years of operation with no corrosion problems occurring. The spoon type bow in conjunction with grounding out the ferry was investigated but a more traditional docking system was utilized due to the shore slope and vertical drop off at the berth. Vince Den Hertog explains the difficulty in maintaining pretension over the run and provides an interesting summary of the variation in the pretensions that occur during the run. The pretension values that were determined allowed for three of the major considerations to be addressed. The pretension controls to a large extent the ferry s transverse deflection due to external forces. Secondly, it provides the driving friction on the bull wheels, and lastly, it is picked sufficiently high so that the tail end tensions never go below zero which would introduce differential loading on the two motors and possible slippage. The shore based facility is fitted with tensioning winches and independent anchors to facilitate proper tensioning and monitoring of cable tension. Mr. den Hertog suggests some physical full scale testing as confirmation which is an excellent idea and will be suggested to BC Ferries. The main traction winches are designed to withstand the breaking loads of the wire. While the forces in this case are significant, it is important to realize that the tension forces of the cable self-cancel within the massive structure of the bull wheel seating and the only forces transmitted to the hull structure are the resistance loads of the ferry plus any variations due to the environmental conditions. The bull wheels and bearing and shafting are designed to withstand the high loads imposed by the breaking load of the cable so that they have a safety factor greater than the cable. Dan McGreer and Chad Oldfield questioned whether the wind loads were determined with vehicles included on the deck. For the purposes of the analysis, the vessel was considered without vehicles on deck since the bulwarks and the central deckhouse presented a substantial windage that would have shielded the vehicles from the athwartships wind conditions. The additional freeboard and more lively motions provided for a situation that we wanted to test. Additionally in the extreme conditions that were tested it was determined that these conditions would represent an emergency crossing with minimal emergency vehicles on board. The reviewers asked if there was a comparison of the viscous damping between the numerical model and tank test modelling. Because of the barge shape of the hull, roll damping will be dominated by wave radiation and likely not viscous effects. The pre-tensioned cables were not included in the verification, however, the viscous damping from the cables in the actual crossing will be significant and the effect is included in the time domain dynamic analysis. The question was asked, The yaw natural period is about 15 seconds. In what loading case is the natural frequency 15 seconds? The yaw natural frequency is 15 seconds in the load case when the ferry is not under wind and wave loading. How does the yaw period change with tension? The yaw period will change with pre-tension of the cables, if you increase the stiffness by 2, you change the natural period 1/sqrt(2). This was all incorporated into the time domain simulations; any issues due to natural frequency being close to wave frequency would show up in time domain analysis. The reviewers asked if the simulation results were sensitive to cable elasticity. The authors respond that the cables axial stiffness will play a role in the dynamic response of the cables to loading. However, in the case of wire rope with its high axial stiffness, the weight of the cable and pretension set with the winches are more critical parameters in the analysis. 312

20 A clarification was asked regarding the fatigue analysis. The dynamic analysis determined a high/medium frequency tension variation due to wind and wave loading. A separate analysis was carried out to determine the fatigue on the cable based on the number crossings made by the ferry. This analysis indicated that while medium/high frequency fluctuations were not a factor, the cyclic loading during the crossing indicated that a yearly switch out of the drive cable to the less onerous guide cable function represented good practice. Alan Winkley asked if this would be the biggest cable ferry ever. We are not aware of any wire rope cable ferries larger than 50 cars, although there is preliminary work being done in Sweden for a larger ferry. There are larger chain ferries in the UK carrying 70 cars. Mr. Winkley asked a number of questions about the cables and any associated hazards. We located the traction winches above deck and along the centerline to minimize maintenance and to maintain symmetry of the drive so the ferry tracked straight. The pretension is done by shore winches which tension the cable and anchors which secure the cables independent of the tensioning winches. Buoys and markers will be used near the berths to indicate the location of the cable. The cable is near the surface less than two boat lengths away from the ferry. Mr. Winkley asked about the operating criteria table. The table was derived by analyzing the wind and wave data to determine when the motions of the ferry of the cable tensions would become limiting. Furthermore, two external criteria were input: the roll should not exceed 5 degrees in normal operations to prevent vehicles sliding on deck and vertical accelerations should not exceed established limits to avoid discomfort to the passengers. RMS and significant values were listed so comparisons to other criteria were obvious. Half the weight of all 3 cables was added to the light ship stability analysis to provide a margin for the weight carried by the ferry in the worst case of loading when the cable is fully picked up off of the seafloor. The weight of the cables is not significant, as the water plane area of the barge is large. Bill Wood made some observations regarding the disadvantages of cable ferries. These were considered by BC Ferries prior to committing to a cable ferry and they felt the advantages outweighed the disadvantages. The cable ferry is being considered for this route for economic reasons, the reduction in crew and reduced fuel consumption of the cable ferry. All prove to be a compelling argument for the cable ferry when the conditions warrant. At Denman Island the relatively benign conditions suit the use of a cable ferry even though the route is perpendicular with the prevailing conditions. A new terminal is being built that addresses the alignment that a cable ferry requires. Mr. Wood asked if a quasi-static method could be used. Preliminary work was performed using Quasi-static methods and showed that three 1.5 diameter cables would be acceptable for the ferry. An increase in the wind loading was subsequently applied and small dynamic fluctuations discovered during the time domain analysis caused us to increase the cable size to 1-5/8 diameter While the quasi- static method has been proven on many small ferries, it was determined that it would be prudent to apply dynamic methods for this unique application since it was larger and longer than previous cable ferries. Mr. Wood asked if BC Ferries was considering the Hornby ferry as well. BC Ferries investigated the use of a cable ferry on the Denman-Hornby route, but found that the route was too exposed to provide a reliable service after assessing the wave conditions. The route crosses Lambert Channel, which is exposed to fetch conditions from the larger Georgia Strait. This is in contrast to the conditions in Baynes Sound. 313