PHYSICAL MODEL TESTS OF ICE PASSAGE AT LOCKS

Ice in the Environment: Proceedings of the 16th IAHR International Symposium on Ice Dunedin, New Zealand, 2nd 6th December 22 International Association of Hydraulic Engineering and Research PHYSICAL MODEL TESTS OF ICE PASSAGE AT LOCKS Andrew M. Tuthill 1 and Carrie M. Vuyovich 1 ABSTRACT A physical model study at CRREL examined the ice passage capability of new lock designs, focusing on important factors such as the configuration of the upper approach, the design of the lock filling and emptying system and the location and design of culvert intakes and outlets. Unconventional ice passage techniques such as manifolds in the miter gates were also evaluated. Physical model results were compared to field observations, and a parallel series of tests using the DynaRICE ice-hydraulic numerical model. This paper describes physical modeling portion of the study and presents preliminary findings. INTRODUCTION Navigation locks are natural collectors of ice and debris, and the problems typically occur when broken (brash) ice, driven by current wind or vessels, collects in the upper lock approach (Tuthill, 22). Vessels then push the ice into the lock chamber, and, in many cases, the ice must be locked through separately, delaying navigation. A lock filling and emptying (FE) system typically consists of culverts in the lock walls that convey flow from intakes located in the upper lock approach, via filling valves, to ports located in or near the lock floor. Many designs have transverse laterals that distribute the flow from the culverts to the ports. The ports typically extend over the central two-thirds of the lock chamber. To empty the lock, flow passes out the ports and culverts, via emptying valves, to outlets in the lower lock approach. To lock ice, the upper miter gates and the lock emptying valves are opened, with the lock at high pool, in order to draw ice into the lock chamber. Often, the surface currents alone cannot move the ice, and a vessel must push the ice about one-third of the way into the chamber then back out and wait while the ice is locked through separately. Once the lock is at low pool, the lower miter gates and the lock filling valves are opened to flush the ice out of the chamber into the lower lock approach. The layout of the upper approach, location of culvert intakes and the design of the FE system all affect how well the lock performs in ice. 1 US Army Cold Regions Research and Engineering Laboratory, 72 Lyme Rd. Hanover, NH 3755, USA

The US Army Corps of Engineers-sponsored Innovations for Navigation Projects (INP) Research Program has developed design concepts aimed at reducing construction costs at new lock projects. One concept significantly reduces concrete volumes by locating the FE culverts in the lock floor rather than their conventional location in the lock walls. Further material and cost savings result from locating the culvert filling intakes directly in front of the upper miter gates rather than the usual location along the walls of the upper approach. Plans for extending existing 183-m-locks to 366 m are also being developed under the INP. One design under consideration would use the FE system of the existing 183-m-lock to fill and empty the extended 366-m-lock. Since these new designs will be used at ice-affected sites, they need to be evaluated in terms their ice passage performance, and the initial design phase is the ideal time to do this. As part of the INP, a physical model study was done in the CRREL 1:36 scale Soo Locks Model to learn how new lock design features influence performance in ice. The physical model results were compared to a parallel set of simulations using the DynaRICE ice-hydraulic computer model (Liu et al., 21), the subject of a second paper. An overall objective of the study was to validate DynaRICE as a tool for the design of ice and debris management measures at navigation locks. PHYSICAL MODELS Tests were done in the 1:36 scale physical model of the Soo Locks in the refrigerated research area at CRREL. Unless noted otherwise, all units for length, time and velocity are converted to prototype using Froude similitude laws. The modeled area included the 33.5 m-wide Poe Lock, which has a split bottom lateral FE system extending over 65 % of the 366-m-long chamber, and the parallel 24.4-m-wide MacArthur Lock with an interlaced bottom lateral FE system covering 74 % of the lock s 289-m-long chamber. Lift is 6.4 m, and the depth in the upper approach and in the lock chamber (at low pool) is about 1 m. For both locks, under existing conditions, the culverts are located in the lock walls and the culvert intakes are in the upper approach walls, just upstream of the upper miter gates. Fig. 1 gives the model layout and Fig. 2 shows a photo. Figure 1: Layout of the CRREL Soo Locks Physical Model. Dimensions are in meters prototype. The study addressed three main ice issues: 1.) Ice accumulation near the culvert intakes during lock filling, 2.) Ability of the FE system to draw ice into the lock chamber, and 3.) Ability of the FE system to flush ice from the chamber. Tests were done in these

three areas in the model Poe Lock and Mac Arthur Locks under existing conditions. The tests were repeated for the Poe Lock with modifications, including through-the-sill intakes located just upstream of the upper miter gates of the Poe Lock, and manifolds in the miter gates to create surface currents to move ice in and out of the lock chamber. Plastic ice A 1:36 scale ice piece size distribution for brash ice was based on field observations made by Daly and Arcone (1989) and analysis of photos of brash ice in the Poe Lock in March of 1999 (Fig. 3). Equivalent full scale diameters for the model ice are: D 15 = 2 cm, D 5 = 65 cm, and D 9 = 2 cm. Crushed polyurethane plastic ice material used in physical model studies by the New York Power Authority (1998) made up the 4 % and finer size fractions, while the larger size fraction ice pieces were sawn from 6.1-mmthick polyethylene. Plastic ice specific gravity is.92. Figure 2: CRREL Soo Locks Model Figure 3: 1:36 scale plastic ice Open Water Velocity Distributions In the quiescent hydraulic conditions found in and around most locks, near-surface water velocity is the most important hydraulic factor influencing brash ice movement. Water velocity in the physical models was measured using drogues, Marsh McBirney electromagnetic probes and acoustic Doppler probes. Two dimensional surface velocity distributions were also obtained through analysis of down-looking video of drifting confetti and plastic ice pieces. The video tracking proved the most useful in unsteady flow situations because it provided progressive horizontal velocity distributions during lock filling and ice lockage operations. Ice Accumulation in the Upper Approach During Lock Filling As an initial condition, the upper approach was filled with a single-layer accumulation of the plastic ice at an estimated surface concentration of about 7 percent and an

average thickness of.46 m. A boom located 98 m upstream of the miter gates kept the ice out of the intake area until the start of the test. Before opening the filling valves, the boom was removed, allowing the ice to drift freely into the intake area. A downlooking video camera tracked the ice movement during lock filling. The lock filling hydrograph, or filling curve, is highly unsteady, increasing from to 18 m 3 /s in the first 3.5 minutes (Fig. 4). During the first half of the hydrograph rise, water currents in the upper approach are more or less depth-averaged. By the time of the hydrograph peak, surface velocity in front of the intakes has significantly declined as an increasing portion of the flow occurs at depth. By mid-fall on the hydrograph, surface currents are practically zero and all flow into the intakes is occurring at depth. Plastic ice movement in the upper approach was observed for 1.) The existing conditions case of 4-m-long intakes located in the sides of the upper approach, beginning 2 m upstream of the closed miter gates, and 2.) Through-the-sill intake case developed in the INP. Both tests used the same filling curve. Fig. 4 shows the discharge into the intakes and compares the position of the leading edge of the ice accumulation with time for the side intakes and the through-the-sill intakes cases. In the side intakes case (1a), the final ice edge position is near the downstream end of the intakes while for the through-the sill case (1b) the ice accumulation reaches the face of the miter gates. 1 2 9 8 Sill Intakes 18 16 Ice Edge Position (m) 7 6 5 4 3 2 Side Intakes Side Intakes Lock Filling Curve 14 12 1 8 6 4 Flow Through Intakes (cms) 1 2 2 4 6 8 1 12 14 16 18 2 Figure 4: Side intake and through-the-sill intakes compared. Drawing Ice into the Lock Chamber For these tests, the lock is at high pool with the upper miter gates open and the lower miter gates closed. Opening the emptying valves 1 % creates surface flow field that decreases from an average surface velocity of.4 m/s in the upper approach to nearly m/s at a location about two thirds of the way down the lock chamber. In the model Poe Lock with the empty valves fully open, total flow is 166 m 3 /s, slightly less than the peak flow during lock filling (Fig. 4).

Initially a.46-m-thick accumulation of plastic ice was retained behind a boom located 98 m upstream of the open upper miter gates. The empty valves were opened, and, once steady flow established, the boom was removed allowing the plastic ice to drift into the lock chamber (Case 2a). In Case 2b, only the upper set of laterals was used, in order to simulate a 183-to-366 m lock extension case being considered in the INP. Flow from the upper laterals was 85 m 3 /s, about half of the Case 2a discharge of 166 m 3 /s. Surface water velocity declined from.3 m/s in the upper approach to zero by about one-third of the way down the chamber. In Case 2c, four 1.1-m-diameter ports in the face of the lower miter gates passed 19 m 3 /s in addition to the 166 m 3 /s passing through the ports in the lock floor. With the ports in the lower gates open, surface velocities in the lower third of the chamber were about.6 m/s. Fig. 5 compares the ice edge positions vs. time for the three cases. The existing FE system performed well, drawing ice to within 6 m of the lower miter gates. Opening the ports in the lower miter gates resulted in only a slight improvement over existing conditions. Using only the upper laterals, it was possible to draw ice about half way into the lock chamber. Distance from Lower Miter Gates (m) 5 45 4 35 3 25 2 15 Upper Miter Gates Upper 1 Lower 5 Lower Miter Gates 5 1 15 2 25 Time (min prot.) Existing Conditions Ports in Low er Miter gates Upper Only Figure 5: Position of ice edge vs. time, drawing ice into the model Poe Lock chamber. Flushing Ice from the Lock Chamber For these tests, the lock is at low pool with the lower miter gates open and the upper miter gates closed. Initially, the entire lock chamber is filled with an ice accumulation of an average thickness of.73 m. Opening the filling valves 1 % results in a total flow of 14 m 3 /s out the ports in the lock floor. The resulting surface flow field increases from near zero at a location about 6 m downstream of the upper miter gates to about.6 m/s in the lower third of the lock chamber. The ice was allowed to drift from the lock chamber into the lower approach (Case 3a). In Case 3b, four 1.1-m-diameter manifolds in the face of the upper miter gates added 19 m 3 /s to the flow from the ports in the lock floor. With the manifolds in the upper gates open, surface velocities in the upstream one-third of the chamber were at or above.3 m/s. Fig. 6 compares the ice edge positions vs. time for the two ice flushing cases, showing a

significant improvement with the ports in the upper miter gates. Distance from Lower Miter Gates (m prot.) 4 35 3 25 2 15 1 5 Upper Miter Gates Upper Low er Low er Miter Gates 2 4 6 8 Time (minutes prot.) Existing Conditions Ports in Upper Miter Gates Figure 6: Ice edge positions vs. time for the two Poe Lock ice flushing cases. MACARTHUR LOCK ICE LOCKAGE TESTS Ice lockage tests were repeated for the MacArthur Lock model to compare ice passage performance for the two different types of filling and emptying systems. The split bottom lateral FE system of the Poe Lock, which extends over 65 % of the locks 366-mlength, consists of two 7-m-long bays separated by 98 m. The MacArthur Lock s interlaced bottom lateral FE system is continuous over 73 % of the lock s 289-m-length (Fig. 1). Fig. 7 compares ice movement into and out of the two lock models with respect to time. Units are normalized by dividing time by total time and location of the downstream edge of the ice accumulation by the length of the lock chamber. The results show no significant difference in performance between the two types of filling and emptying systems, although the ice accumulation moves faster initially in the Mac Arthur Lock. SUMMARY AND CONCLUSIONS Physical models were used to investigate the relationship between lock design features and ice passage performance. Topics investigated include the effect of culvert intake design on ice movement in the upper approach and the relationship between filling and emptying system design and ice lockage performance. During lock filling, the through-the-sill type intakes drew ice into contact with the upper miter gates while the conventional side intakes the maximum ice position was about 3 m away. Based on this result, at sites where heavy ice or debris is expected, the cost advantage of the through-the-sill intake should be weighed against possible operational

Distance from Lower Miter Gates/ Lock Length 1.4 Drawing Ice Into Poe 1.2 Flushing Ice from Poe Drawing Ice Into MacArthur 1. Upper Miter Gates Flushing Ice from MacArthur.8 Poe Upper.6 MacArthur.4 Poe Lower.2 Lower Miter Gates...1.2.3.4.5.6.7.8.9 1. Time / Total Time Figure 7: Ice lockage performance of Poe and MacArthur Locks compared. disadvantages. Addition of manifolds in the lower miter gates produced a very slight improvement in terms of drawing ice into the chamber. The addition of ports in the upper miter gates significantly improved the model lock s ice flushing capability. Observations of a fullscale ice flushing ports at the Soo Poe Lock support the model results. Based on limited model testing, there appears to be no significant difference between the ice lockage performance of the The MacArthur and Poe Lock models in spite of their differing filling and emptying system designs. REFERENCES Daly, S.F. and Arcone, S.A. Airborne Radar Survey of a Brash Ice Jam in the St. Clair River. CRREL Report 89-2, Cold Regions Research and Engineering Laboratory (1989). Liu, L., Tuthill, A.M. and Shen, H.T. Dynamic simulation of ice conditions in the vicinity of locks and dams. In Proceedings: 11 th Workshop on River Ice, University of Ottawa, Ottawa, Ontario, Canada (21) 145 153. New York Power Authority. Hydropower on the Upper Niagara River; Studies of Measures to Mitigate the Impacts of Ice on Power Generation and Shoreline Property. Niagara Power Project, FERC No. 2216-NY, Niagara Falls, New York, March 31 (1998). Tuthill, A.M. Ice Affected Components of Locks and Dams. U.S. Army Corps of Engineers ERDC/CRREL Technical Report 2-4 (22).