Physical Modeling to Develop Approach Velocities at Trashrack Intakes

By Andrew Craig, Troy Lyons, Skyler Street and Duncan Hay

This article has been evaluated and edited in accordance with reviews conducted by two or more professionals who have relevant expertise. These peer reviewers judge manuscripts for technical accuracy, usefulness, and overall importance within the hydroelectric industry.

The 955.6-MW Priest Rapids hydroelectric project on the Columbia River in Washington is owned and operated by Grant County Public Utility District (PUD) No. 2. A concrete gravity arch dam 178 ft in height by 10,103 ft in length impounds the Columbia River, creating the 237,100 acre-ft Priest Rapids Lake. The scheme also includes a powerhouse that has 10 units, a 22-bay spillway and two fish ladders on the left and right side of the project.

A study was conducted from 2011 to 2013 in support of Grant County PUD’s plan to install new pumps in the left bank fish facility and to ensure hydraulic conditions associated with the larger pumps would meet criteria for pump intakes established by the National Marine Fisheries Service.

Pump flows vary based on the plant’s tailwater elevation. The left bank fish ladder is supplied with water through a gravity intake gate from the forebay and five 500-hp pumps that draw water from the tailrace. Grant County PUD planned to add two 700-hp pumps to the existing pump station to increase attraction water supply to the fish ladder and decrease spill by way of the gravity intake gate.

 

[bc_video account_id=”1214147015″ player_id=”Hkvuzcnml” video_id=”4996940328001″ min_width=”480px”]

Physical models

This study used two physical models: a 1:64 scale Priest Rapids tailrace model and a 1:12 scale Priest Rapids pump station model.

Tailrace model

The 1:64 scale tailrace model was used to estimate the far-field velocity/sweeping flow conditions, with modifications to achieve more accurate conditions near the left bank fish ladder. The pump station face wall was updated to create representative pier protrusions to more accurately represent flow conditions near the wall and pump inlets. The bathymetry near the pump station was also updated with higher- resolution data from a 2011 multi-beam sonar survey and dive survey data.

Pump station model

The 1:12 scale pump station model was constructed in IIHR’s Environmental Flow Facility – a closed-loop, recirculating-type flume.

The flume is equipped with two variable frequency drive (VFD) motors that recirculate a fixed volume of water. The flow passes through a contraction and baffles at the upstream end and into the test section. The test section is 65 ft long by 7.5 ft high and normally 10 ft wide, which was reduced to 51.7 in wide with a partition wall for this study. The drive motors are capable of a maximum 120 cubic feet per second (cfs) through the test section.

The pump bay model was installed 37 ft downstream of the contraction. At the downstream end of the channel, a tailgate is installed to aid in setting the tailwater elevation.

Field conditions were replicated in the sectional model by setting sweeping velocities along the trashrack face, which were verified with a SonTek Acoustic Doppler Velocimeter (ADV).

The modeled pump bay was equipped with a series of three pressure taps along the upstream and downstream side walls near the lower inlet.

Direct scaling of the prototype trashrack resulted in a bar thickness that was structurally unfeasible in the model. To overcome this issue, an equivalent model trashrack was used for a case of low flow (410 cfs), low sill elevation (372.67 ft) and tailwater sufficient to submerge the entire screen.

The left image shows the location of the Priest Rapids facility in Washington, and the right image indicates the location of the study.
The left image shows the location of the Priest Rapids facility in Washington, and the right image indicates the location of the study.

Vertical bar width was scaled from the prototype by 1:12 and then tripled in thickness to provide rigidity in the model.

There are nine windows (openings) between framing members and analysis was conducted for one window of the screen setup.

There are three identical screens fabricated from aluminum plate that have vertical bars, which are sized and arranged so head losses are replicated correctly. Horizontal backing structures on each unit create three windows for fluid to pass through.

The modeled trashrack was designed to accurately replicate the prototype trashrack head losses.1

A VFD-controlled motor and pump combination mounted above the pump bay generated discharges up to and exceeding prototype intake flows of 500 cfs.

The velocity field at the entrance of the model trashrack was measured using a two-component TSI laser Doppler velocimeter (LDV) system.

A two-dimensional (2D) automated traversing system was installed in the overhead bay, downstream and adjacent to the pump bay. The traverses were electrically connected to a two-axis controller, which received move commands from data acquisition software. Limit switches were incorporated on each axis to safeguard the traverse payload from interfering with the facility or model.

The 1:64 Priest Rapids tailrace model featured a spillway with the powerhouse on the right.
The 1:64 Priest Rapids tailrace model featured a spillway with the powerhouse on the right.

The measurement system consists of a 5-W argon-ion laser beam separator fiber bundle that has four transmitting fibers, one receiving fiber, a 15-mm stainless steel underwater LDV probe, a two-channel LDV signal processor and a computer running TSI Flowsizer software for data acquisition.

The 15-mm probe has a nominal 120-mm focal length in air, but the underwater application extends this to 159.6 mm due to the index of refraction of water.

The transmitted beams were folded 90 degrees upward by an IIHR-designed and built stainless mirror module, as well as a laser mirror to enable measurement of beam path shapes in the reference (V) and sample (N) configurations or VN (green beams; 514 nm) and sweeping velocity (blue beams; 488 nm) in the model coordinate system.

Once the beams exit the probe, the beam path travels through the folding mirror to the measurement volume where the measurement is made. This enables measurement of the normal velocity component at the trashrack entrance in a non- intrusive manner.

From a library of predetermined measurement area grid positions for semi- automated surveying, software coordinated data acquisition cycles in conjunction with traverse moves. Working from a home position in the lower left corner of the measurement area, surveys consisted of a series of 286 to 364 acquire-and-move cycles. The LDV probe samples data across a typical row and from 22 to 28 rows depending on the tailwater setting for the tests, which sets the extent of the free surface and the vertical size of the measurement grid.

The probe then moved up to the next row and worked across in the opposite direction. This created a zig-zag pattern for the data acquisition path. Data was sampled in random mode with 20,000 samples acquired over both channels before advancing to the next grid point. Because the data is acquired in random mode, no turbulence information is available in the archival dataset.

1:64 scale tailrace model velocity measurements

Tests were undertaken to calibrate model velocities to field measurements provided by Grant County PUD. The test configuration in Figure 1 (this page) depicts the baseline velocities and compliant velocity.

Velocity measurements were obtained at 20%, 60% and 80% depths at various locations along the pump house face using a 2D, side-looking SonTek ADV probe. For Test 1 flow conditions, velocities were measured 5-ft-prototype away from the pump house face at the center of pump bay units Nos. 1 and 7. For Test 2 flow conditions, velocities were measured 9.5-ft-prototype away from the pump house face at the center of pier bays Nos. 3 and 4.

The modified bathymetry inside and outside of the excavated area was painted, and to provide bed roughness, sand in varying amounts was applied until satisfactory results were obtained for use during the study.

Sweeping velocity measurements

A series of experiments were conducted on the 1:64 scale tailrace model to determine the maximum sweeping velocities in front of the left bank pump house, during varied conditions, to provide input for testing on the 1:12 pump station model of a single left bank pump house bay. To achieve conservative values of sweeping velocities, the pump house was not in operation during this series of tests.

The 1:64 Priest Rapids tailrace model featured a spillway with the powerhouse on the left bank fish facility pump house.
The 1:64 Priest Rapids tailrace model featured a spillway with the powerhouse on the left bank fish facility pump house.

A three-dimensional, down-looking micro ADV SonTek probe was used to take measurements at the center of pump house bays Nos. 1 and 2 at 10-ft-prototype from the pump house face. Velocities were measured at 50 Hz rate for 1 minute and time-averaged.

1:12 scale pump station model tests

The objectives of the 1:12 scale pump station model studies were to develop and document a pattern of backer plates on the trashrack of a pump bay that will house a new pump to verify that the normal velocities near the trashrack face are equal to or less than 1 ft/sec (fps) and to ensure acceptable head losses from the tailrace to the pump chamber.

The tests were conducted to map the mean velocity field a distance of 6-in-prototype from the face of the pump bay trashrack and measure head loss across the screen. Only velocity components normal to and sweeping across the screen, VN and sweeping velocity, respectively, were of interest in this study.

This photo shows the 1:12 scale pump station model with the TSI laser Doppler velocimeter setup at the trashrack face.
This photo shows the 1:12 scale pump station model with the TSI laser Doppler velocimeter setup at the trashrack face.

Several flow conditions and pump bay geometries were iteratively studied to help understand how changes in these variables affect flow physics through the trashrack screen.

Flow condition variables included tailwater elevation, pump discharge, and sweeping velocity. Pump bay geometry variables included the extent to which the pump bay headwall was extended and trashrack backer plate porosity.

Test designs and conditions

Phase 0 tests were designated to determine optimum LDV data acquisition settings, procedures, LDV probe configuration, standoff distance from probe to trashrack face, measurement grid densities and locations, and backerplate porosities and their effects on VN and compliance with VN<1 ft/sec criterion.

For these tests, tailwater = 404 ft, sweeping velocity = 3.91 ft/sec and pump discharge = 380 cfs were held constant as these flow conditions were established and recognized as starting points for the archival data acquisition in subsequent phases of the experiment.

The laser Doppler velocimeter system shown here is taking velocity measurements at the model trashrack face. Visit http://www.hydroworld.com and search for the article-related video: 'Use of Physical Models.'
The laser Doppler velocimeter system shown here is taking velocity measurements at the model trashrack face. Visit hydroworld.com and search for the article-related video: “Use of Physical Models.”

Phase 1 initiated the archival data collection and, in this case, the headwall was intact. For Phase 2, the headwall was split at 409.67 ft and the lower section was removed. But tailwater, sweeping velocity and pump discharge remained the same as Phase 1 tests. For Phase 3, the shortened headwall was retained but tailwater, sweeping velocity and pump discharge were all increased to investigate more extreme initial conditions. In this phase, and in addition to balancing the flow at the trashrack screens, the effect of systematically reducing tailwater from 411 ft to 407 ft was investigated.

For Phase 1 to 3 testing, a baseline test was completed without using backerplates. Then, VN was balanced to comply as closely as possible with VN <1 ft/sec by installing fixed-porosity backerplates on the three screens and statistically monitoring VN across the measurement area. The backerplates were fabricated from high-density polyethylene sheets on a CNC router using 0.25-in tooling to create symmetric, regular hole patterns behind each of the screen’s three windows. The number of holes and spacing were laid out with computer-aided design software to obtain the desired porosity in each of the screen windows.

Phase 1

Phase 1 tests consist of five surveys on a measurement grid to achieve acceptable VN compliance. This creates a relatively narrow inlet at the lower part of the pump bay for flow to enter.

Baseline results show gross velocity is a reference velocity through the trashrack based on a completely open area and pump discharge. For this case, compliance is 67.8%. VN is very strong in the lowest trashrack where there is a pathway through the inlet not blocked by the headwall. Sweeping velocity is high at the lower leading edge of the pump bay but decelerates across the pump bay face when VN increases and enters the inlet.

The end result of balancing VN for Phase 1 tests shows compliance is 98.3% after addition of backerplates with relatively closed porosities (7.5%, 7.5%, 7.5%, 7.5%, 10.0%, 12.5% and 12.5%) for the bottom seven windows, which produces three small regions near the leading edge of screen No. 1 where VN >1 ft/sec. To achieve this flow condition, head loss through the screens is increased one order of magnitude, from 0.192 ft to 2.076 ft.

Phase 2

Phase 2 tests consist of nine surveys to achieve acceptable VN compliance. This creates a larger inlet across the pump bay entrance for the flow to enter.

The test configuration figure shows baseline velocities depicted at left, and the compliant velocity field conditions depicted on the right.
The test configuration figure shows baseline velocities depicted at left, and the compliant velocity field conditions depicted on the right.

Baseline results for this case indicate compliance is 71.0%. VN is strongest and non-compliant in the lower right corner of the pump bay. Sweeping velocity is strongest in the lower left corner of the pump bay but diffuses away from this location due to the effects of the flow turning into and entering the pump bay.

Phase 3

Phase 3 tests consist of three surveys to achieve acceptable VN compliance, plus another seven surveys to investigate effects of tailwater and sweeping velocity separately. For this case, pump discharge is increased from 380 cfs to 450 cfs. The pump bay geometry for Phase 2 tests is also used for Phase 3 tests.

Baseline results for this case show compliance is 74.2%. Similar to Phase 2 baseline results, VN is strongest and non-compliant in the lower right corner of the pump bay. Sweeping velocity is strongest in the lower left corner of the pump bay but diffuses away from this location due to the effects of the flow turning into and entering the pump bay.

The end result of balancing VN for Phase 3 tests shows compliance is 100% after addition of backerplates with porosities (17.5%, 17.5%, 17.5%, 17.5%, 20.0%, 22.5%, 22.5%, 22.5% and 22.5%) for all nine windows from bottom to top. To achieve this flow condition, head loss through the screens is increased 3.8 times, from 0.132 ft to 0.504 ft.

Summary and conclusions

Three phases of testing were undertaken, each beginning with the as-built trashrack in place to establish baseline intake velocity and head loss conditions. Each phase of testing incorporated different river tailwater, pump bay discharges and/or pump bay geometry.

In each phase of testing, baseline conditions were used to inform researchers where areas of noncompliant velocity zones were located and to subsequently design and install backer plate arrangements for testing. This iterative process was repeated until a compliant velocity field and head loss conditions were achieved.

The findings of this study can be used in operational logic of the future pumps to limit flows and maintain inlet trashrack criteria. This has not yet been implemented at Priest Rapids Dam.

Note

1Weber, L.J., M.P. Cherian, M.E. Allen and M. Muste, “Headloss Characteristics for Perforated Plates and Flat Bar Screens,” IIHR Technical Report, No. 411, IIHR-Hydroscience & Engineering, Iowa City, Iowa, 2002.

Reference

Craig, A., et al, “The Use of Physical Models to Balance Fine Trashrack Intake Approach Velocities in a Complex Flow Field,” Proceedings of HydroVision International 2015.

Andrew Craig, P.E., and Troy Lyons, P.E., are engineers at the University of Iowa’s College of Engineering IIHR Hydroscience & Engineering. Skyler Street is a plant mechanical engineer at the 1,040-MW Wanapum hydro project. Duncan Hay is principal at Oakwood Consulting Inc.

Previous articlePreparing for the Worst
Next articleTech Briefs

No posts to display