Examination of an Electrified Bar Rack Fish Guidance Device for Hydropower Turbines

Abstract

The potential of hydropower turbines to cause injury or mortality to fish is a concern. To reduce this risk, engineers have begun to develop a conically arranged, cantilevered electrified bar rack (Center Sender). This device is proposed to be mounted within the intake of a turbine, guiding downstream-passing fish towards the center of the turbine where blade velocities are lower and blades are thicker, likely reducing the potential for fish to be injured. A simplified version was installed in a flume for testing with Rainbow trout (Oncorhynchus mykiss) while examining several parameters such as electrification, bar spacing, bar angle, and water velocity. The most effective settings were observed to be a water velocity of 1.0 m s⁻¹ with all bars installed at an angle of 40° with electrification on. Other combinations were still effective but had varying results. A minimal proportion (2.3%) of fish passed at lower velocities with bars electrified and mounted at 20°, suggesting that while it functions well as a guidance device in high-velocity conditions, it performs better as a barrier at lower velocities. The study suggests that the Center Sender has promising potential for reducing the harmful effects of hydropower turbines on fish populations by guiding them away from stressors. Further research is needed, but current results encourage its potential use in hydropower operations.

1. Introduction

Fish guidance is a common practice to divert fish for many different purposes, such as invasive species control, preventing fish from entering intakes at water resource facilities (irrigation, cooling systems, hydropower, etc.), and guiding fish away from various other dangers. Coarse trash racks, and sometimes fine fish screens, are commonly installed on the intakes of hydropower turbines to keep fish and debris from entering the turbine. If fish pass through turbines, they may be exposed to several stressors that can cause injury or mortality [1]. Therefore, it is common practice, and often required, to install some sort of fish diversion at hydropower facilities.

Screening and guidance technologies, based on a variety of operation principles, have been developed to prevent fish from entering hazardous areas of hydropower facilities and other water infrastructure [2,3,4], as well as for the exclusion of invasive species [5]. Physical barriers block fish from entering hazardous zones by preventing a passage exclusively by reduced bar spacing [6]. However, the fine spacing required to exclude small fish creates debris and ice-clogging issues and often requires costly ongoing operation and maintenance [7,8]. Additionally, if water velocities exceed the swimming abilities of fish, impingement resulting in injury or mortality can occur [4].

Behavioral barriers are intended to divert fish that may be small enough to not be physically excluded from intakes or other hazardous areas. Techniques include bubble curtains, sound, light, and electric fields [3,9,10]. Mechanical barriers, which are intended to influence the behavior of the fish by creating hydraulic cues such as turbulent flow zones, high local velocity gradients, or rack parallel flow are another type of behavioral barrier [11,12]. The success of behavioral barriers has been mixed and is often species-dependent [10].

The application of electricity has mostly been used as a deterrent, forming behavioral barriers to keep invasive fish species out of certain areas such as hydropower outlets [10,13,14]. Electrical barriers function by passing electricity from an anode to a cathode through water, which creates an electrical field in the surrounding area [15]. When exposed to this electrical field, fish generally exhibit avoidance behavior [5].

It may be possible to exploit this avoidance behavior and use electricity to guide fish in a desired direction, and this application has been investigated by several research institutions in recent years [16,17,18,19,20]. The FishProtector-Technology developed by the Department of Hydraulic Engineering at the University of Innsbruck (Austria) combines both a mechanical and a behavioral barrier (electric field in the water) to create a hybrid barrier [21]. Downstream migrating fish tend to follow the main flow towards the turbine intakes at hydropower plants [22], visually perceive the mechanical barrier, and react by orienting themselves rheotactically positively (facing upstream) [23]. While further approaching the barrier, they enter the electric field emitted by the electrodes and eventually react with quick bursts of upstream swimming [23]. Thus, the passage of the barrier is reduced [18,23]. The mechanical barrier itself can either be horizontally tensioned steel cables (Flexible FishProtector) which also function as electrodes [21] or classical bar racks equipped with additional electrodes mounted on the upstream face of the bars (Bar Rack FishProtector) [24]. Reactions to the electric field depend strongly on factors such as pulse parameters and voltage, the electric conductivity of the water, species, size, and physical conditions of the individual fish [10,13,14,25,26]. If the electric field parameters exceed a safe range, fish can experience severe injuries [27].

This study examined the potential for an electrical guidance system to be used at a hydropower facility to guide fish to safer passage routes. There may be potential to install guidance equipment just upstream of the turbine, which may guide fish through the expected safest area of the turbine and reduce the chances of injury or mortality for fish. For many of the propeller type turbines (i.e., Kaplan, Francis, bulb, etc.), the center of the rotor, or hub section, is often designed with thicker blade widths and, due to the radial nature, has a lower velocity than regions of the turbine closer to the blade tips [28]. These features make the occurrence of blade strikes much lower, and, if a blade strike does occur, it is less likely to be injurious than strikes that occur near the blade tips. If fish can be successfully guided to the center of the rotor, there may be a significant increase in the survival of fish passing through the turbine.

Turbine intakes have substantially higher velocities (generally between 1 and 4 m s⁻¹) than regions further upstream (i.e., reservoir, headwaters, etc.), and electrical barriers have been found effective only at low velocities (0.5 m s⁻¹ or below) [3,4,29]. The purpose of this research was to determine if a cantilevered, electrified bar rack module could effectively guide fish at a flow velocity of 1 m s⁻¹, representative of a turbine intake, as well as a lower velocity representative of previous studies, 0.4 m s⁻¹. The electrified bar rack concept was submitted to the 2020 US Department of Energy and US Bureau of Reclamation’s “Fish Protection Prize” under the name “Center Sender”, referring to the use case of guiding fish to the typically safer center region of a rotating turbine. In this application, the Center Sender is a conical-shaped electrified bar rack that can be mounted just upstream of the guide vanes of a turbine to direct entrained fish to the center of the rotor—near the hub—where strike speeds are slowest. For this study, a single cantilevered bar rack module consisting of six bars spaced 60 mm apart was designed and installed in a specialized flume and assessed using Rainbow trout (Oncorhynchus mykiss).

2. Materials and Methods

2.1. Test Flume

The test flume was built at the Natel Energy facility in Alameda, California. It was constructed of transparent acrylic with a steel frame and consisted of several connected sections, including a 4.88 m-long (16 ft) enclosed section (Modules 1 and 2; 0.65 m × 0.65 m square) which was reduced to a 0.4 m × 0.65 m section with a trapezoidal prism insert (Figure 1). Bar rack modules could be installed or removed externally on the flume wall as illustrated in Figure 1. Open-air sections on either end of the closed flume section allowed for fish to be inserted in and removed from the flume. Removable sliding screens allowed the open-air sections to be separated from the enclosed section, such that fish could be confined to the accessible areas during acclimation and retrieval from the flume. Flow velocities up to 1 m s⁻¹ within the 0.4 m × 0.65 m confined section were provided by an ABS J 604 ND submersible sump pump and conditioned by two Cheng rotation vanes (CRVs) at the upstream elbows, as well as a 2.3 m long diffuser (Figure 1).

2.2. Flow Velocity Measurement

The actual velocity distribution in the reduced flume section at the two operating velocities was measured with acoustic doppler velocimetry (ADV) using a SonTek FlowTracker2 handheld Acoustic Doppler Velocimeter. The transects consisted of 15 samples each (three depths, five lateral locations) and were taken at a section 1 m upstream of the bar rack mounting location, and 1 m downstream (sample points denoted by the encircled ‘x’ symbols in Figure 1). The measurements were conducted without the bar rack module in place, and the probe was inserted from above into the closed flume section through sealed ports to maintain closed-flume conditions.

2.3. Cantilevered Bar Rack Modules

Two cantilevered bar rack modules were tested: one inclined at 20° to the flow, and the other inclined at 40° to the flow (

Table 1. Physical parameters of the cantilevered bar rack modules.

Parameters	20° Module	40° Module
Number of bars	6	6
Bar diameter	8 mm	10 mm
Bar length	1023 mm	545 mm
Bar spacing	60 mm	60 mm
Insulated region (projected distance from base wall)	50–200 mm	32–129 mm

). The modules consisted of six spiral-wrapped stainless-steel round bars spaced 60 mm apart and clamped on one end within an insulated plastic base. The bar of both modules extended 350 mm into the projected cross-section in flow direction from the orthographic right-hand side of the flume (i.e., the clear spacing between the far wall and the tip of the bars was held constant at 300 mm for both modules). Spiral wrapping (0.2 mm diameter) was applied to the round bar to minimize flow-induced oscillations.

In preliminary testing, it was observed that some fish tended to approach the module from downstream after their initial passage and would enter the acute angle region formed by the bar rack module and the flume wall. Trapped in this confined space, these fish tended to contact and pass back through the bar clearances for a prolonged period of time, which caused visible stunning and injuries when the bars were electrified, and current was transferred through the fish body. To prevent these injuries, the ends of the bars adjacent to the base (in the acute angle region) were insulated using electrical tape.

The Center Sender module bases accommodating the round bars were CNC machined from polyurethane MB-0670 (Rampf GmbH & Co. KG, Grafenberg, Germany). The bases consisted of two parts, with slots for the bars, bolted together to tightly clamp the cantilevers. Internal cable ducts allowed for the connection of the bars and wires connected to the pulse generator unit. The 2.5 mm² cables were routed to the pulse generator via the shortest way.

A prototype pulse generator unit (University of Innsbruck, Austria) was used for the generation of the pulsed current for electrification of the CS module. The used pulse parameters were low voltage square wave pulsed DC (≤80 V). This minimized the potential of fish injuries while reducing power consumption and maximizing deterrence and guiding effects [13,15]. The necessary amperage needed for fish guidance depends on several factors including the used voltage, surface area of electrodes, number of electrodes, conductivity of the water, etc. In the performed experiments, the maximum amperage varied between 35–40 mA. The generator was supplied by an EA PSI-8080-40 DT (EA Elektro-Automatik GmbH & Co. KG, Viersen, Germany) laboratory power supply and created the gated burst current pulses necessary for the deterrence and guiding effect of the behavioral barrier. The pulses in the range of milliseconds were arranged in sets of five pulses, with longer pauses between sets of several hundred milliseconds. This specific pulse pattern has been demonstrated as effective and harmless in prior investigations [15,16,18].

2.4. Fish

Rainbow trout, ranging in fork length from 169 to 286 mm (mean = 229 mm; Figure 2), were acquired from Smith’s Trout Farm, in Calistoga, California, and transported to the Natel Energy facility in Alameda, California. Rainbow trout were selected for several reasons including their representation of salmonids which many populations are threatened and migrate making them susceptible to encountering hydropower facilities and other water infrastructure. Additionally, rainbow trout are easily obtainable and have been extensively studied in previous research. Fish were held in a four-tank setup, which consisted of three independent systems: two 1136-L (300-gal) circular polyethylene tanks sharing a single pump with a water turnover rate of 7.2 min, and two other identical tanks on individual pumps with water turnover rates of 4.3 min each.

2.5. Water Quality

Water quality measurements were taken prior to each trial in the holding tank and in the flume. During testing, temperatures within the flume were on average 13.0 °C and averaged 12.2 °C within the holding tanks, representative of summer river conditions for rainbow trout native habitat (

Table 2. Water quality measurements recorded throughout the testing period.

Water	Temperature (°C)			Dissolved Oxygen (mg L−1)			Conductivity (µs cm−1)
	Mean	Min	Max	Mean	Min	Max	Mean	Min	Max
Holding tanks	12.2	10.0	14.4	11.0	10.3	11.7	325	290	360
Flume	13.0	8.2	16.3	10.7	9.2	11.9	306	201	320

). Water conductivity within the flume ranged from 201 to 320 µs cm⁻¹ during testing, representing a common range observed in freshwater rivers within the United States [30].

2.6. Testing Procedure

Prior to testing, the flume was set to the desired testing parameters. Fourteen different combinations of water velocity, bar angle, voltage, and electrification pattern were tested (

Table 3. Various parameter combinations tested in the flume.

Water Velocity (m s−1)	Bar Incline Angle	Module Configuration	Voltage (V)
0.4	20°	All Bars	0
0.4	20°	All Bars	0
0.4	20°	All Bars	75.5
0.4	20°	Top and Bottom	0
0.4	20°	Top and Bottom	75.5
0.4	40°	All Bars	0
0.4	40°	All Bars	80.0
0.4	40°	Top and Bottom	0
0.4	40°	Top and Bottom	76.5
0.4	N/A 1	None 1	0
1.0	20°	All Bars	0
1.0	20°	All Bars	75.5
1.0	20°	Top and Bottom	0
1.0	20°	Top and Bottom	75.5
1.0	40°	All Bars	0
1.0	40°	All Bars	80.0
1.0	40°	Top and Bottom	0
1.0	40°	Top and Bottom	76.5
1.0	N/A 1	None 1	0

Note: ¹ Controls were conducted for both water velocities without the Center Sender module installed in the flume.

). Each replicate was initiated by capturing ten fish with a dip net from the stock holding tanks and placing them in buckets to be transported to the insertion module of the flume (Figure 1). Fish were removed from the buckets and placed in the screened insertion area of the flume and allowed to acclimate for two minutes at 0 m s⁻¹. For the 0.4 m s⁻¹ tests, the velocity was then increased to 0.4 m s⁻¹ for an additional two minutes. For the 1 m s⁻¹ tests, the velocity was increased from 0 m s⁻¹ to 0.4 m s⁻¹ for one minute, and then to 1.0 m s⁻¹ for an additional minute. From the preliminary testing, it was determined that four minutes was sufficient for fish to acclimate and exhibit normal swimming behavior and this applied to all replicates.

After the acclimation period, the screen was removed from the insertion section and fish were allowed to move downstream to encounter the cantilevered bar rack module. During this period, fish were recorded using two cameras (GoPro Hero 9 and 10) mounted above and to the side of the bar rack device. The trial was conducted for ten minutes or was ended once all fish passed through the flume. Once fish passed the cantilevered bar rack module and entered the downstream fish access chamber, they were recaptured via a long-handled dip net. The fish were then placed in buckets and transported to an observation area where each individual was evaluated for injury (i.e., discoloration due to exposure to an electrical field, abnormal swimming behaviors, or wounds) and measured. Injury rates were recorded per replicate, but observations were not linked to individual fish. Once all fish were examined and measured, they were placed in buckets and transported back to a holding tank where they were further monitored for 48 h. After the 48-h evaluation, the fish were euthanized. Each fish was an independent sample and was not used in additional trials.

2.7. Video Analysis

Video analysis was conducted using the video that was captured from above the Center Sender module. The flume area was separated into seven lanes, each 100 mm wide, except for lane seven which was 50 mm wide (Figure 3). Two crosswise sections, representing approach and passage, were identified for each of the three bar incline angles that were examined (control, 20°, and 40°; Figure 3). The approach location was the lane in which the fish entered the proximity of the bar rack module device. This location was just outside of the observed detection range where fish began to respond to the electrical field and also corresponded with a flange on the flume which facilitated a consistent landmark for the analysis. The passage location was the lane at which the fish passed the threshold of the bars, beginning at the tips of the bars and proceeding to the side of the flume, perpendicular to the flow (Figure 3). Two types of events were recorded for fish that crossed the approach location: aborted—fish that swam back upstream, or passage—fish that passed the bar rack module or control passage location.

2.8. Statistical Analysis

The effect of test day was evaluated to ensure any differences observed in passage position were not due to differential fish behavior between days. Passage positions of the control group (flow velocity = 0.4, 1.0 m s⁻¹; module angle = none; electrification mode = none) from each day were first tested for normality using the Shapiro-Wilk normality test (α = 0.05). If normally distributed, passage positions of control fish by day were tested for homogeneity of variance using Levene’s test (α = 0.05). If variances were homogenous between days, passage positions of control fish by test day were compared using a t-test (α = 0.05). If variances were not homogenous between days, passage positions of control groups were compared using Welch’s test (α = 0.05). If control group passage positions were not normally distributed, the nonparametric Mann-Whitney-Wilcoxon test was used to evaluate differences between days.

Assuming no difference in passage positions of control fish between test days, multiple linear regression modeling was used to evaluate the factors affecting passage position, which served as the response variable in the model. Categorical predictor variables included flow velocity (0.4 or 1.0 m s⁻¹), module angle (N/A, 20°, or 40°), and electrification mode (none, top and bottom, or all), and the two-way interactions of flow velocity × module angle and flow velocity × electrification mode. Stepwise forward multiple regression was performed on the response variable, passage position. Factors were added at a significance level of p < 0.10. Model selection was conducted using F-tests. Goodness-of-fit of the selected model was examined by plotting observed passage positions versus model-predicted passage positions.

3. Results

Testing was conducted from 11 through to 14 April 2022.

3.1. Flow Uniformity

ADV measurements of the flume showed a maximum deviation of 5.0% to 9.1%, with maximum velocity occurring at the bottom nearest to the mounting wall for the bar rack module (

Table 4. Velocity measurements taken by acoustic doppler velocimetry for the 0.4 m s⁻¹ condition, 1 m upstream and 1 m downstream of the bar rack module mounting location.

	Lateral Position X (mm)
		0.4 m s−1 Condition, Upstream						0.4 m s−1 Condition, Downstream
		108.3	216.7	325.0	433.3	541.7		108.3	216.7	325.0	433.3	541.7
Depth Y (mm)	100	0.365	0.360	0.361	0.405	0.424	100	0.376	0.374	0.377	0.381	0.376
	200	0.362	0.381	0.376	0.3890	0.419	200	0.372	0.374	0.384	0.396	0.415
	300	0.392	0.394	0.400	0.413	0.427	300	0.406	0.400	0.412	0.409	0.419

and

Table 5. Velocity measurements taken by acoustic doppler velocimetry for the 1.0 m s⁻¹ condition, 1 m upstream and 1 m downstream of the bar rack module mounting location.

	Lateral Position X (mm)
		1.0 m s−1 Condition, Upstream						1.0 m s−1 Condition, Downstream
		108.3	216.7	325.0	433.3	541.7		108.3	216.7	325.0	433.3	541.7
Depth Y (mm)	100	0.990	0.941	0.936	0.982	1.033	100	0.995	0.999	0.970	0.965	0.953
	200	0.988	0.952	0.975	1.020	1.087	200	1.018	1.005	0.995	0.990	1.025
	300	1.009	1.032	1.039	1.0657	1.098	300	1.017	1.046	1.065	1.060	1.069

). Leakage flow through the trapezoidal insert averaged 8.4%.

3.2. Fish Approach and Passage

Passage positions of control fish were not normally distributed on 3 of 4 days (for the 3 that deviated from normality, Shapiro-Wilk p ≤ 0.032, and for the 1 that was normal S-W p = 0.185). Variances were not homogeneous (Levene’s p = 0.017), and passage positions were similar (Welch’s p = 0.154).

Throughout testing, a total of 801 approaches were observed in the downstream direction, 433 of which resulted in the passage of the cantilevered bar rack module (

Table 6. The number and rates of approach and passage events of fish encountering the cantilevered bar rack module for each treatment condition. The column with the heading “Bars” refers to the module configuration and the number of bars installed, including no bars for controls (None), top and bottom bars only (T&B), and all six bars (All).

Water Velocity (m s−1)	Bar Incline Angle	Bars	Voltage (V)	Trials	Number of Fish	Number of Approaches	Approaches per Fish	Number of Passages	Passages per Fish 1	Passage per Approach Rate
0.4	N/A	None	0	7	70	78	1.11	72	1.03	92.3%
0.4	20	T&B	0	2	20	25	1.25	21	1.05	84.0%
0.4	20	T&B	75.5	2	20	67	3.35	17	0.85	25.4%
0.4	20	All	0	3	30	22	0.73	16	0.53	72.7%
0.4	20	All	75.5	2	20	128	6.40	3	0.15	2.3%
0.4	40	T&B	0	2	20	9	0.45	8	0.40	88.9%
0.4	40	T&B	76.5	2	20	100	5.00	20	1.00	20.0%
0.4	40	All	0	2	20	22	1.10	20	1.00	90.9%
0.4	40	All	80.0	2	20	45	2.25	18	0.90	40.0%
1.0	N/A	None	0	8	80	86	1.08	79	0.99	91.9%
1.0	20	T&B	0	2	20	26	1.30	25	1.25	96.2%
1.0	20	T&B	75.5	2	20	37	1.85	17	0.85	45.9%
1.0	20	All	0	2	20	20	1.00	19	0.95	95.0%
1.0	20	All	75.5	2	20	38	1.90	13	0.65	34.2%
1.0	40	T&B	0	2	20	24	1.20	24	1.20	100.0%
1.0	40	T&B	76.5	2	20	25	1.25	21	1.05	84.0%
1.0	40	All	0	2	20	21	1.05	21	1.05	100.0%
1.0	40	All	80	2	20	28	1.40	19	0.95	67.9%
		Combined		48	480	801	1.67	433	0.90	54.1%

Note: ¹ Several fish were observed to have multiple passage events because after initially passing the cantilevered bar rack module, the fish reascended the flume and subsequently descended again, passing the module more than once.

). The approach lane was more evenly distributed across the flume lanes for testing conducted at 0.4 m s⁻¹ (Figure 4) compared to 1.0 m s⁻¹ (Figure 5) which had distributions skewed towards the outer lanes, particularly lane 7. Additionally, an aborted passage was more common at 0.4 m s⁻¹, particularly for trials conducted with electrified bars and especially for the treatment with all bars set at 20° inclination; this treatment had the most approaches and also the fewest passage events (Table 6).

3.3. Regression Analysis

Plotting the passage lane of control fish as a function of their entrance lane revealed a significant correlation with the entrance lane explaining 43% of the variability in the passage lane (

Table 7. Results of a simple linear relationship of passage lane as a function of entrance lane for control fish.

Term	Estimate	SE	t Ratio	Prob > |t|	Lower 95%	Upper 95%
Intercept	2.556	0.268	9.540	<0.001	2.027	3.086
Entrance location	0.533	0.050	10.570	<0.001	0.433	0.632

). Both the intercept and slope terms were significant with 95% confidence intervals that did not contain 0 and 1, respectively, indicating the effect of the entrance lane on the passage lane did not follow the 1:1 line. The observed relationship indicated control fish that entered closer to lane 1 were more likely to move toward the center of the flume than those that entered closer to lane 7. Therefore, the difference between the passage lane and entrance lane (i.e., lane movement) was used as the predictor variable in all analyses to represent the distance that fish moved between the entrance and passage.

Using analysis of covariance (ANCOVA) to fit lane movement (i.e., passage lane minus entrance lane) as a function of test day + entrance lane + test day × entrance lane indicated no significant effect of test day (F ratio = 2.177; p = 0.094) or the interaction of test day × entrance lane (F ratio = 1.721; p = 0.165), suggesting that the behavior of control fish did not differ by test day. Therefore, it was assumed that the behavior of treatment fish also did not differ by test day, allowing data from all test days to be pooled for further analyses.

ANCOVA was used as a screening tool to evaluate the effect of each predictor variable (water velocity, angle, number of bars, and voltage) on the relationship between entrance lane and lane movement (i.e., passage lane minus entrance lane) for control and treatment fish combined. All main effects and interaction terms were significant (F ratio > 5.096; p < 0.025). These correlations indicated that fish were more likely to move away from the side of the flume to which the bar rack device was mounted at the higher velocity, when voltage was on, and when the bar rack module was present in the flume, particularly when deployed at a 40° angle with all six bars installed (Figure 6).

Multiple linear regression modeling was used to more closely evaluate the relative effect of predictor variables on the lane movement of treatment fish. Categorical predictor variables included flow velocity (0.4 or 1.0 m s⁻¹), module angle (20° or 40°), number of bars (two or six), and electrification mode (voltage off or on). Entrance lane was included as a continuous predictor variable. All two-way interactions between predictor variables were also included as candidates in the model selection procedure. Stepwise forward multiple regression was performed on the response variable, lane movement (i.e., passage lane minus entrance lane). Factors were added at a significance level of p < 0.10. Model selection was conducted using F-tests. Goodness-of-fit of the selected model was examined by plotting observed passage positions versus model-predicted passage positions.

The final model provided a good fit to the observed data, explaining 70% of the variability in lane movement and included all main effects, interactions of entrance lane with velocity, number of bars, and voltage, and interaction between velocity × angle (

Table 8. Results of a multivariable linear regression model of lane movement (i.e., passage lane minus approach lane) fit as a function of entrance lane, water velocity, angle, number of bars, and voltage for treatment fish exposed to the cantilevered bar rack module.

Term	Estimate	SE	t Ratio	Prob > |t|	Lower 95%	Upper 95%
Intercept	4.900	0.181	27.13	<0.001	4.544	5.255
Velocity [0.4]	−0.283	0.071	−3.97	<0.001	−0.423	−0.143
Angle [20]	−0.141	0.069	−2.03	0.043	−0.277	−0.004
Bars [2]	−0.054	0.068	−0.79	0.430	−0.188	0.080
Voltage [Off]	−0.161	0.067	−2.38	0.018	−0.293	−0.028
Approach lane	−0.797	0.034	−23.57	<0.001	−0.863	−0.730
Velocity [0.4] × Angle [20]	−0.175	0.069	−2.55	0.012	−0.311	−0.040
Velocity [0.4] × (approach lane − 5.004)	0.088	0.035	2.55	0.011	0.020	0.156
Bars [2] × (approach lane − 5.004)	0.089	0.033	2.73	0.007	0.025	0.154
Voltage [Off] × (approach lane − 5.004)	0.099	0.032	3.13	0.002	0.037	0.161

). The model indicated that at the lower velocity tested (0.4 m s⁻¹), the cantilevered bar rack module was not effective at guiding fish when deployed with top and bottom bars only at an angle of 20° without electrification (Figure 7). However, the device was effective at redirecting fish under all other scenarios. At the lower velocity, fish guidance was more effective with all six bars instead of top and bottom only, at 40° instead of 20°, and with voltage on instead of off (Figure 7). At the higher velocity tested (1.0 m s⁻¹), the angle of the bar rack had little effect on fish guidance (Figure 8). However, all six bars were more effective than the top and bottom bars only, and electrification on was more effective than electrification off. The cantilevered bar rack module was most effective at a flume velocity of 1.0 m s⁻¹, with all six bars present, electrification on, and with the bars positioned at an angle of either 20° or 40°.

Only slight descaling was observed on two fish, one of which was a control (water velocity 1.0 m s⁻¹), and neither could be attributed to the bar rack device. The only other observed injury was hyperpigmentation, which was present in fish that contacted the rack bars while they were electrified (Figure 9). This phenomenon was observed on 32 fish total and occurred more frequently for fish tested at a water velocity of 0.4 m s⁻¹. Twenty-three occurrences of hyperpigmentation were observed for the 0.4 m s⁻¹ “electrification on” cohort (29%) compared to nine occurrences for the 1.0 m s⁻¹ “electrification on” cohort (11%). The hyperpigmentation was observed to fade within the first hour after exposure to the electrified bars and fully dissipated within 24 h. No mortality was observed during testing or the post-testing holding period.

4. Discussion

The results of this study suggest that the cantilevered bar rack module was effective at guiding fish to the desired area, with minimal injury to the fish. Based on the parameters examined, the most effective settings for the device were a water velocity of 1.0 m s⁻¹ and all six bars installed at an angle of 40° with electrification on. While these were the settings that performed the best, several other combinations were also effective. This suggests that a cantilevered bar rack device, or conical Center Sender module arrangement, has the potential to guide fish to safer areas within a hydropower turbine environment or other water infrastructure.

This study examined fish over a small size range (95% of the tested fish were between 200 to 250 mm); however, fish of different sizes are known to respond differently to an electric field. For example, during electrofishing, smaller fish require a higher field intensity and threshold current to be immobilized compared to larger fish [31,32]. Smaller fish may be less likely to be guided by an electrified bar rack device; however, smaller fish are also less likely to be injured by blade strikes when passing through a turbine [33]. Therefore, a threshold of fish size may be determined for which a “Center Sender” style cantilevered electrified bar rack would be most effective. Additionally, the response to electricity can vary among different species, sizes, and life stages and may also be influenced by water temperature, water chemistry, and time of day or season [25]. The operating parameters of the electrified bar rack device could be tuned for maximum effectiveness based on the expected sizes and species of fish present at a facility during different times of the year.

A minimal proportion of fish (2.3%) passed the Center Sender during the trials conducted at a water velocity of 0.4 m s⁻¹ with all bars electrified and mounted at 20°, indicating that the electrified bar rack performs better as a barrier than a guidance device at low velocities. While turbine intake velocities typically exceed 0.4 m s⁻¹, this velocity may be more representative of diversions or irrigation intakes, and the barrier function of the electrified bar rack is consistent with results reported for similar apparatuses designed to function as barriers [16,18,20,23,34].

This study provides a proof of concept for the cantilevered, electrified bar rack device to function as a guidance system for fish passing through turbines. Additional research is needed to identify the optimal operating conditions and physical configuration of the device and to confirm that guided fish would follow streamlines and pass near the center of the turbine, at lower velocities, and with lower mortality rates as a consequence. Future research should focus on expanding the size range of tested fish and examining additional species, as well as a field study of the device for guidance and barrier applications. Overall, the cantilevered, electrified bar rack module effectively guided Rainbow trout within a flume setting of 1.0 m s⁻¹ and shows promise for guiding fish in high-velocity conditions such as turbine intakes and other water infrastructure.