Why cylindrical screens in the Columbia River (USA) entrain few fish

Abstract

Cooling-water intakes often entrain large numbers of larval and juvenile fish, leading to questions why a water intake on the Columbia River, USA, with axial flow, cylindrical screens on a T-screen structure entrained very few (four fish in 4,007 x 10³ m³ of water sampled over four years). This is despite being downstream of spawning areas for Chinook salmon Oncorhynchus tshawytscha releasing up to 56 million pre-smolts and having screens consisting of porous plate with 9.5-mm-diameter pores capable of passing salmon fry and other small fish (poor physical exclusion). Multiple publications of experiments and observations in hydraulics and biology are used to help explain the low entrainment, concluding that effects of the screening structure rather than the actual screen are most important. Primary mechanisms appear to be: (1) bow-wave-like hydraulics at the structure’s nose cone that deflect small fish away from the screen’s pores, and (2) fish detection and avoidance of pressure and velocity changes upstream of the structure that aid deflection. These combined hydraulic and biological-behavior effects of the structure would cause salmon fry and other small fish to bypass the screened portion of the intake. At the screens, (3) high ratio of sweeping flow to approach and through-screen velocities likely further prevents entrainment. Although site- and design-dependent, this analysis may be useful for evaluating or planning other screen installations and focusing regulatory attention on hydraulics of the structure as well as the screen.

Keywords:

• Entrainment
• fish
• salmon
• screen
• cylindrical
• forensic ecology

Introduction

Entrainment of aquatic organisms is an important concern when natural waters are withdrawn for use. The amount of water for thermal-electric power plants and diversions for agricultural irrigation is often large enough that significant numbers of fish and other aquatic organisms can be affected. Traditionally, the entrainment risks have been minimized by design criteria for intake screens that have focused on pore or slot sizes and approach and through-screen water velocities (Weisberg et al. 1987; Bell 1991; USBR 2006; Karchesky and McDonald 2007; NMFS 2011; EPA 2014). Pore or slot sizes are intended to be small enough that target organisms cannot physically pass through. Approach and through-screen velocities are intended to be low enough that the target organisms can swim away from the screen openings despite the inflow.

For example, the U.S. National Marine Fisheries Service, Portland District (NMFS), which regulates intake screens in the Columbia River, USA, requires circular screen-face openings where anadromous fish are present to not exceed 2.4 mm in diameter and the approach velocity to not exceed 0.12 ms⁻¹ for active screens, or 0.06 ms⁻¹ for passive screens (NMFS 2011, Section 11, pages 94 and 89, respectively). These criteria were developed for rotating drum screens, vertical screens, and inclined screens typical for water intakes in the U.S.’s Pacific Northwest (NMFS 2011, page 89).

Experience has shown, however, that actual entrainment reduction is more complex. As Turnpenny and O’Keeffe (2005) noted, ‘Fish screens usually form only part of an overall fish diversion or protection system and it is the performance of the entire system that needs to be proven.’ Noting the large number of techniques now in use, however, they envision ‘an almost unlimited agenda for testing in order to cover the different environments, species, and life stages.’ They note further that, ‘In addition to any generic research needs, site-specific commissioning trials may also be required [for innovative technologies] to show that screening measures perform satisfactorily.’

Performance of the 40-year-old T-screen intake system at Energy Northwest’s Columbia Generating Station (CGS) near Richland, Washington, USA (described below) provides a useful example of a successful screening system in a potentially risky situation for a valuable fish stock. Although not part of a systematic, comprehensive testing program envisioned by Turnpenny and O’Keeffe (2005), this intake system provided an opportunity for a site-specific performance evaluation of the ‘overall protection system’ using their ‘generic’ type of information. During re-permitting of the station’s cooling-water system, concerns were raised that the axial-flow cylindrical screen system used at CGS did not meet the NMFS screen criteria for pore size despite being operated for nearly 40 years without apparent fish entrainment. The cooling-water intake for CGS was designed and constructed before there were agency screen guidelines, and is somewhat unique. Nonetheless, the selected design of the screening system was intended to minimize risks to Chinook salmon (Oncorhynchus tshawytscha) and steelhead (O. mykiss) which spawn and initially rear in the 53-km-long Hanford Reach of free-flowing Columbia River immediately upstream.

To evaluate the fish exclusion performance of the T-screen intake, a fish-sampling system was installed in the shoreline pumphouse and initially operated in pre-operation (1979) and post-start-up (1980) periods (Mudge et al. 1981; WPPSS 1985). The power station pumps river water (for cooling-tower make-up) from an on-shore pump well that is connected to the intake screens in the river; water enters the screens by gravity flow from the river to the pump well. Two metal cages can be lowered to completely cover each of two 0.91-m-diameter intake pipes and collect any entrained fish during 12- or 24-hour monitoring periods. The capture efficiencies of the cages were determined by inserting a known number of identifiable fish of entrainable size in the cages, holding in the monitoring position for a typical monitoring period, retrieving the cage and recounting. Capture efficiencies were 80% in 1979 and 1980 (some test fish escaped the cages through insufficiently tight seals).

The entrainment studies were repeated in 2018 and 2019, with capture efficiencies of 90.5% in refurbished cages (Energy Northwest 2020). The new studies also quantified the main fish resource at risk of entrainment. Chinook salmon reproduction in the Hanford Reach has been monitored in recent years. In 2018, the estimated number of pre-smolts produced and passing CGS ranged between 10.7 million and 56.4 million using different estimation protocols; in 2019 the estimates were between 8.1 million and 29.9 million (Energy Northwest 2020). Pre-smolts occupy both nearshore waters (resting, feeding) and channel flows where the CGS intake screens are located (migration) (Dauble et al. 1989). Highest risk of entrainment would be during night-time migration (22:00 to 04:00 hours; Dauble et al. 1989).

Despite abundant and vulnerable early life stages of salmon, steelhead and other small fishes in the river in the spring (summarized by Energy Northwest 2020) there were only occasional individual salmon or other small fish captured in monitoring studies. In the pre-operational (1979) monitoring study, no fish were captured in 625x10³ m³ of water sampled. In 1980, no fish were captured in 470x10³m³ of water sampled. In 2018, 1.53⁻⁰⁶ fish per m³ were captured from 1,477.8x10³ m³ of water sampled in circa 25-hr periods for 15 days April-September (one Chinook salmon and one Pacific lamprey Lampetra tridentata); in 2019 1.61⁻⁰⁶ fish per m³ were captured in 1,434.0x10³ m³ of water sampled in circa 25 hr periods for 15 days March-September (one unidentifiable salmon and one torrent sculpin Cottus rhotheus).

One wonders what features of the CGS intake screen system account for its remarkably low entrainment. Could these features inform further development and deployment of intake-screening systems elsewhere? There are many types of intake screens and screening systems, with valuable performance evaluations (e.g., Hanson 1979; Heuer and Tomljanovich 1979; Weisberg et al. 1987; Turnpenny and O’Keeffe 2005; USBR 2006). The focus of these evaluations was nearly always on passage through the screens per se. Assessments of some other operating cylindrical screens indicated that entrainment of fish eggs and larvae by those screens appeared to be less than could be explained by physical exclusion by screen pore or slot size alone (Weisberg et al. 1987; Enercon 2010). The CGS screening system is different from those studied earlier and may yield further insights through a site-specific forensic ecology study (sensu Willard 1980) using generic information.

This paper employs pertinent, generic scientific information on the fluid dynamics of cylindrical objects in flowing water and fish behavior when encountering an in-stream obstacle to help explain the documented safe passage around the CGS screening system. The emphasis is on functional hydraulic and biological principles for fish protection that differentiate between effects of the CGS intake system and the actual screens. The information in this forensic review served as a prelude to detailed Computational Fluid Dynamics (CFD) modeling of the CGS screen system.

CGS cylindrical screen system

The CGS screen system is typical of axial-flow, cylindrical T-screens designed for flowing water (Mudge et al. 1981). It consists of two 9.1-m-long, 1-m-diameter, horizontal cylinders with solid cones at the ends of a metal frame that supports the porous screens forming much of the cylinder. The central vertical exit pipe gives the entire structure a T shape (Figure 1). The CGS’s two T-screen structures are placed with long axes oriented nearly parallel to each other and to the mean flow direction of the river.

Figure 1. A one-meter diameter, 9.1 m long, cylindrical T screen unit used at the Columbia Generating Station on the Columbia River near Richland, Washington. Screen areas are stippled. The leg of the T carries water to the power station. River flow is from left to right. (profile view; courtesy of Energy Northwest).

Each screen structure has two cylindrical screens consisting of 2-m-long sections of 1.07-m-diameter perforated pipe having 9.5-mm-diameter holes covering about 40% of the area (the perforated-pipe screen material preceded the adoption of wedge-wire in cylindrical screens evaluated by Turnpenny and O’Keeffe 2005). The screen assembly includes a 0.91-m-diameter inner pipe with 19.1-mm-diameter holes covering 7% of the area to provide an even distribution of water velocities through the outer fish-protection screen. Water enters the screen perpendicular to ambient water flow and is withdrawn from the central cavity of the inner cylinder through the exit pipe. Initial hydraulic modeling of the intake screen system concluded that at maximum water withdrawal rates of 1.6 m³s⁻¹ the maximum through-screen velocity (through a pore at the external screen surface) would be approximately 0.2 ms⁻¹ and the approach velocity vector perpendicular to the screen face 2.5 cm from the outer screen surface would be 0.03 ms⁻¹ (through-screen and approach velocities are standard measures for evaluating fish exclusion (NMFS 2011)) No screen-cleaning system has been used at the CGS because of low debris loads in the river and insignificant growths of algae on the structure seen in diver inspections.

The screen material in similar cylindrical screen systems has consisted of perforated plate (Alam et al. 1974) or, more recently, wedgewire (Heuer and Tomljanovich 1979; Weisberg et al. 1987; NAI and ASA 2011a). Slot sizes of 2 to 9 mm have been used for installed wedgewire screens (NAI and ASA 2011a) but slot sizes can go down to 1 mm (Weisberg et al. 1987).

River environment

The screen units are placed in an unimpounded reach of the Columbia River at River Kilometer 566 (46.477967, −119.265305; bing.com), approximately 8 km upstream of the city limits of Richland, Washington, USA. The units lie approximately 90 m west of the center of the river channel in the 274-m-wide river (June elevation; Energy Northwest 2020). The intakes are approximately 1.2 km downstream of an island where the main channel shifts from the east side of the island toward the western shore.

The screens operate at seasonally changing river velocities and depths during the months when salmon and steelhead juveniles are present (March-October). For the historical record from 1917 through 2011, mean monthly velocities in April, June and October were 1.5, 1.8 and 1.2 ms⁻¹ and water depths of 4.8, 6.0, and 3.0 m, respectively (Energy Northwest 2020).

Entrainable fishes

The Hanford Reach of the Columbia River, where the CGS is located, hosts 21 fish species, ranging from uncommon to abundant (Energy Northwest 2020). Some are year-around residents, occupying mainly backwaters or benthic habitats, and some are seasonal migrants. Most prominent are salmonids. The fall race of Chinook salmon spawns (October and November) and rears (mid-March to mid-June) in the Hanford Reach with the early juveniles particularly abundant and vulnerable to entrainment since they occupy both shallow and deep water (Dauble et al. 1989). As noted in the introduction, their numbers can be tens of millions of individuals in a season. All of them that survive the early rearing season must pass the vicinity of the CGS intake during outmigration. Steelhead also spawn in the Hanford Reach, with some historical redds (nests) located within 1 km upstream of the CGS intakes. Their juveniles are most abundant in deep water. Other salmon found in the Hanford Reach occur only as migrants: upstream migrating adults and downstream smolts (spring race of Chinook salmon, coho salmon O. kisutch, and sockeye salmon O. nerka).

Although salmonids of all sizes pass the CGS’s water intake, only the smallest are at potential risk of entrainment. Most juvenile salmon migrating or foraging in the mid-Columbia River are much larger than the CGS’s pore size (Dauble et al. 1989; Bell 1991) and therefore not at risk. Emergent Chinook salmon and steelhead alevins are about 25 mm long; steelhead fry and Chinook salmon fry at the stage when the yolk sac disappears that could be rearing in the vicinity of CGS are about 35 mm long; Chinook Salmon zero- age foragers are about 50 mm; Chinook salmon sub-yearling migrants are about 75 mm; Chinook salmon yearling migrants are about 100 mm; wild steelhead pre-smolt are about 125 mm; and hatchery steelhead are about 150 mm at release. Thus, the most appropriate concern is for the Chinook salmon and steelhead alevins that are newly emerged from spawning areas and their early fry.

Forensic analysis

Although the river at the CGS water intake has abundant fish life, with species and life stages vulnerable to entrainment as measured by the screen pore sizes, the CGS monitoring studies show this does not equate to actual entrainment. Determining why this is so could aid design and placement of screening systems elsewhere. Ideally, this forensic analysis would seek hydraulic and fish behavior data from around the actual screen system. Unfortunately, the river is at seasonal flood stage when the vulnerable fish are present and in situ data collection is unsafe. There are, however, much relevant data in the literature on hydraulics and fish behavior (‘generic studies’ of Turnpenny and O’Keeffe) to bring to bear for developing an understanding why so few fish have been entrained. It is useful to take a step-by-step view of a potential entrainment event, considering in turn strictly hydraulic mechanisms, fish avoidance stimulated by hydraulic features, and hydraulics at the screen pores.

Hydraulic mechanisms

Juvenile salmon or steelhead vulnerable to entrainment would be moving downstream (during migration or after displacement from shallow habitats) in the river water that is travelling 1.5 to 2 ms⁻¹ at the screen system. Some would be moving in a trajectory that meets a T-screen system. No other route to a screen is plausible, such as swimming from downstream. River velocities are clearly higher than the swimming abilities of salmon fry, which have a 3-min swimming speed capability of only 0.183 to 0.244 ms⁻¹ for fish 34 to 40 mm long (Easterbrooks 1984).

A fish would first encounter the fluid dynamics around the nose cone of CGS’s cylindrical screen structures. The general pattern of such fluid flow around an obstacle is well understood from several fields, including aeronautics (Seiff and Whiting 1962), submarine design (Amiri et al. 2018), open channel flow (Mignot and Riviere 2010), fish swimming mechanics (Liao 2007), and prey escape behavior (Stewart et al. 2014). Waves from the bows of ships were actually drawn by Leonardo da Vinci five centuries ago (Raissi et al. 2020). The early study by Seiff and Whiting (1962) compared hydraulics from several bow profiles, including a shape similar to the nose cones of the CGS screen structures.

Application of these dynamics to the CGS screen structure is shown conceptually in Figure 2. Upstream of the nose cone, ambient velocity is rapidly reduced, and pressure increased, by the presence of the obstacle (conceptually denoted by the P = P_ambient isobar in Figure 2; note that in reality there would be a smooth, asymptotic departure from ambient pressure). There follows a zone of low to zero velocity and high pressure immediately upstream of the nose cone (the stagnation region). The spatial extent of the perturbed flow field depends on the magnitude of the ambient velocity and the geometry of the obstacle.

Figure 2. Conceptual view of flow streamlines and altered pressure and velocity around the upstream end of a cylindrical screen structure in a river, illustrating hydraulics of a ‘bow wave’ that would keep fish from entering a screen either hydraulically or by avoidance of pressure/velocity changes. Drawing by B. Mater.

These velocity and pressure changes and the resulting flow around the obstacle result in the hydraulic phenomenon commonly referred to as a ‘bow wave’ in general reference to the lateral divergence of flow around the bow of a ship. In the case of a surface-penetrating obstacle like a ship (or a near-surface obstacle), the increase in pressure at the leading end of the obstacle manifests as an elevation of the free surface – a standing gravity wave in the reference frame of the ship. For a submerged obstacle, the free surface perturbation caused by the high pressure region decreases with depth of submergence, but the high pressure region nonetheless persists as a standing pressure wave near the leading end of the obstacle. Particles such as fish, with inertia, tend to be deflected more than the water, as labeled. As such, the term ‘bow wave’ has historical precedence as a general descriptor and remains useful for this forensic analysis. Although usually portrayed in two dimensions, the hydrodynamic signature of the bow wave is three dimensional for a deeply suspended cylinder with its axis parallel with the water flow.

There is reasonable generic evidence that a small fish drifting toward the nose cone would be deflected by the bow wave and not follow water flow lines into the CGS screen. As water flow lines wrap around the structure, inertia would cause particles to diverge outwardly from stream lines just downstream of the nose cone. The empirical evidence comes from rigorous generic studies of entrainment of early life stages of fish into model T-screen cylindrical structures in axial flow laboratory flumes at the Alden Research Laboratory (NAI and ASA 2011a, 2011b). The objective was to quantify entrainment of fish eggs and larvae through wedgewire T-screens that might be deployed in the Hudson River USA for condenser cooling water at the Indian Point Energy Center (IPEC, owned and operated by Entergy Corporation). A 2010 study (NAI and ASA 2011a) tested scale models of cylindrical T-screen structures with combinations of four screen slot sizes (2, 3, 6, and 9 mm), two through-screen velocities (7.62 and 15.24 cms⁻¹), and placed in water flowing at three velocities (7.62, 15.24, and 30.48 cms⁻¹). Neutrally buoyant beads, zebrafish (Danio rerio) and white sucker (Catastomus commersoni) eggs, alive and dead larval Atlantic tomcod (Microgadus tomcod), striped bass (Morone saxatilis), white sucker, and hybrid striped bass x white bass (Morone chrysops) (224,068 total; overall 0.3-23 mm long) were added to the water upstream of the screen structure. Tested species were selected as size and shape surrogates for the variety of entrainable species in the Hudson River. Day and night tests were run. Test subjects were collected with nets after entrainment through the screen, carried past the screen, and excluded from (retained on) the screen. The trials were also videotaped. Several tests were repeated in 2011 in which higher ambient flow velocities were tested (45.72 and 60.96 cms⁻¹), fluid dynamics of the system was modeled (Gessler and Laumbay 2012), and the CFD models were used to plot trajectories of water, inanimate beads (simulating fish eggs), killed organisms (having appropriate shape and mass but unable to exhibit behavior), and live organisms (capable of behavior) as they either passed the screen or entered it (NAI and ASA 2011b). Data collected from entrainment and in the several nets were used to develop statistical models (combined for all species) of the fates of test organisms.

These landmark studies identified three hydraulic mechanisms as contributing directly to reducing the percentage of injected beads, eggs or larvae that were entrained through the screen:

• Hydraulic bypass by the bow wave dominated the fate of fish in most trials. The bow wave caused physical ejection of the particle away from the nose cone of the screen structure. Forces attributed to inertia tended to further move drifting materials away from the screen openings at the circumference of the screen structure, even though water trajectories entered the screen’s slots. Larger larvae and juveniles exhibited greater deflection from the screen structure, even when not alive.

• Exclusion by the slot size and through-screen velocity, the traditional measures of entrainment risk, was a relatively minor contributor to fish fate.

• Sweep off by near-screen sweeping flow, involving eggs and larvae initially impinged on the wedgewire screen, moved eggs and larvae progressively along the screen and eventually washed them off. This, too, was a minor factor.

Performance of a surrogate screen system was also tested in the Hudson River (2 mm slot width, 7.6 cms⁻¹ through-screen velocity; ASA Analysis and Communication et al. 2012) to validate the laboratory flume results. The estuarine river had variable tidal flow rates. Although the field conditions did not allow detailed particle tracking, the concentration of organisms entrained through the screen was significantly less than the concentration measured in the adjacent river water. The test validated the Alden flume studies (NAI and ASA 2011a, 2011b) and provided even better entrainment prevention than the laboratory flume studies.

One notable difference between the studies for IPEC and the CGS intake system is the much higher ambient velocities in the Columbia River than in either the test flume or the Hudson River. Nonetheless, the broad applicability of fluid dynamics at obstacles to multiple media and physical structures and the hydraulics of a bow wave indicate that this difference should not preclude relevance. If anything, the higher ambient velocities in the Columbia River should amplify the hydraulic patterns seen in the flume studies and bypass more fish hydraulically. The 2011 flume studies showed this trend for the range of velocities they studied (the fraction not entrained increased for a given organism length for ambient velocities from 7.62 cms⁻¹ to 60.96 cms⁻¹; NAI and ASA 2011b). Also, Hanson (1979, as cited in Turnpenny and O’Keeffe 2005) showed decreased entrainment as channel velocities increased from 1.5 cms⁻¹ to 3 or 6 cms⁻¹.

Sweeping (or transport) flow is another hydraulic factor. Sweeping flow is the velocity parallel with the screen face in contrast to the through-screen velocity (through the pores) and the approach velocity (the velocity toward a pore at a designated distance from the screen surface) (Rajaratnam et al. 2006). Sweeping-velocity is to be greater than the approach velocity to avoid entrainment or impingement of fish on the screen (NMFS 2011). The flume studies for IPEC measured a ‘sweep off’ of fish and inanimate objects at the screen surface as another minor factor that reduced entrainment, which they attributed to the sweeping flow (NAI and ASA 2011a, 2011b).

The river’s sweeping flows at the CGS intake (monthly average 1.5 to 1.8 ms⁻¹ during Chinook salmon migration) are well above the intake’s calculated maximum through-screen velocity (approximately 0.2 ms⁻¹) and approach velocity (0.03 ms⁻¹ at 2.5 cm from the outer screen surface) (Mudge et al. 1981). The sweeping velocity likely differs, however, depending on where screen pores are located on the screen structure. The upstream screen will likely be affected by the bow wave such that there is a lower sweeping velocity there whereas the sweeping flow may be near the river velocity at the downstream screen (due to streamlines folding back after the bow wave; Figure 2). Nonetheless, sweeping flows that exceed the approach and through-screen velocities by wide margins through much of the screen area likely contribute to the monitored near lack of entrainment.

Overall, much of the available generic evidence points to hydraulic features of the CGS intake structures, especially hydraulic bypass by the nose cone’s bow wave, as a major contributor to near non-existent entrainment at the CGS intake. As the IPEC flume studies showed, hydraulic bypass at the screen structure’s nose cone is largely the same for inanimate or biological particles, species or life stage, alive or dead. Sweeping-flow velocities much higher than through-screen and approach velocities likely make entrainment through the screen unlikely.

Detection and avoidance by fish

Despite empirical generic evidence that fluid dynamics alone at the nose of the CGS intake system may be responsible for negligible entrainment, behavioral avoidance is likely involved also. It is useful to view the dynamics at screen systems, including at CGS, from the perspective of the fish (Coutant 1999). Swimming or drifting fish, even in larval stages, are known to detect and avoid obstacles, which is stimulated by changes in water pressure and flow. Fish have a lateral line system of water-filled tubes in their skin for detecting water flow and pressure in their surroundings (Bleckmann 2007). The system is connected to the surrounding water and contains cells that detect water movement within the tubes. Most prominent along each side of a fish (thus called the ‘lateral line’ system) it also occurs in the head. It often is the location for formation of air bubbles in gas-supersaturated water in Columbia River basin salmon (Coutant and Genoway 1968; Dawley and Ebel 1975). The lateral line system appears early in the development of fish (Blaxter 1986), and is well developed by the time salmonid alevins such as those in the Hanford Reach emerge from spawning gravels.

A fish uses its lateral line system to create a pressure and water-movement ‘picture’ of the surroundings to identify and avoid nearby physical objects, an ability termed hydrodynamic imaging (Hassan 1989). Hassan demonstrated this with blind cave fish, but the mechanism is also used by sighted fishes. In addition to sensing nearby objects by their pressure and velocity signals, a swimming fish generates a flow field around itself due to water displacements at the head and tail regions (Teyke 1988; Hassan 1989; Windsor et al. 2010a, 2010b). The ambient pressure and velocity fields provide a ‘base case’ of the fish moving in water. When the fish moves near an obstacle, its flow field is distorted in a characteristic way that depends on the obstacle’s geometry and dimensions. For a moving obstacle, or one in flowing water, the fish’s stimulus distortion depends the interaction of the fish’s flow field with the flow field around the object as well as the object’s physical location (Hassan 1993). The spatial distributions of the pressure/velocity flow fields have been derived mathematically as well as conceptually (Hassan 1992, 1993). The lateral line system also detects water velocity and direction characteristic of turbulent eddies, which are ubiquitous features of a flowing river environment. This likely aids migration or positioning in turbulent flows (Liao 2006).

The pressure and velocity changes at the nose cones of the CGS’s intake systems likely initiate behavioral avoidance. Because a fish’s hydrodynamic imaging ceases when it stops swimming (Teyke 1985), juvenile Chinook salmon migrating downstream past the CGS intake are likely slowly swimming head upstream while drifting and responsive to the pressure and velocity changes upstream of the CGS intake’s nose cones. Swimming into the current is a common orientation of Chinook salmon fry in the wild in the Hanford Reach and in laboratory tanks (personal observations). It is generally understood that this behavior provides stability control and rapid responsiveness to environmental stimuli (Coutant and Whitney 2000; Liao et al. 2003). A slow, head-upstream swimming orientation provides opportunity for a longer warning time when encountering an obstacle than when swimming with the current facing downstream (which some fish species do, at least sporadically). Since downstream migration of juvenile Chinook salmon occurs mostly at night (Dauble et al. 1989), hydrodynamic imaging must replace vision for detecting an object.

Likely avoidance of the CGS intake by salmon fry detecting pressure and velocity changes is supported by studies of predator avoidance (Stewart et al. 2014). In that case, the cylindrical object is a moving predator fish approaching a stationary larval fish, its prey (the hydraulics are of interest here, not the predation). The nose of the advancing predator creates a hydraulic pattern similar to that seen at a fixed obstacle in flowing water (Figure 2). In laboratory experiments with high-speed video, larval fish were seen to detect the change in pressure/velocity ahead of the predator and initiate an instinctive startle response. The startle response consists of darting behavior involving a rapid tail flip consisting of a fast coiling of the body into a C shape followed by rapid extension into an L shape, which propels the fish away from the initial location (Taylor and McPhail 1985). The fish in those experiments may escape being eaten or, as applied to an intake screen structure, flip away from the nose cone and into the hydraulic bypass of the bow wave.

The Alden flume studies for IPEC identified behavioral avoidance of the test screen structure by the fish larvae in most test runs (NAI and ASA 2011a, 2011b). The larvae apparently were able to sense the flow field perturbations of the bow wave at the nose of the model screen structure and actively move away from the screen sufficiently to generally avoid entrainment. Some larvae may have detected the screen structure visually (an accessory cue), although tests conducted at night were consistent with day tests. Larvae also showed avoidance reactions when they were drawn close to the screen’s slots. As with hydraulic bypass, larger larvae showed the most behavioral avoidance (fish larvae longer than 20 mm showed 90% probability of avoiding entrainment, likely due to increased swimming ability of larger larvae). Behavioral avoidance was also shown in field studies of a model screen system when larvae actively avoided an unscreened control port that was initially intended to serve as the reference intake for ambient concentrations (ASA and NAI 2012).

Personal in-river observations provide additional evidence for fish detection of an obstacle upstream of the physical structure. Downstream movements of juvenile steelhead at night were followed as they approached an experimental flow deflector and shoreline features (Bevelhimer and Coutant 2008). The hatchery-produced steelhead (∼ 21 cm long) moved downstream with water flow into a power plant intake canal on the St. Joseph River, Michigan. The fish, marked by chemoluminescent light tags, drifted head upstream in the downstream flow and showed a clear ability to detect the obstacles about a half meter upstream of them. Of the 61 tagged fish that approached the deflector from upstream, 39 (64%) went around it by passing widely to the side or under it. The remainder slowed their movement and often held position in the velocity stagnation region immediately in front before moving through slots in the array, a behavior discussed by Liao (2007).

Therefore, there is sufficient generic evidence to support fish behavioral detection of a high-pressure region at the upstream end of the CGS’s screening structures (nose cones) that can instigate a rapid avoidance response by fish as they drift toward the unit. This, in combination with altered flows (due to the loosely termed ‘bow wave’) could yield the distinctly low entrainment rates at the CGS.

Discussion and conclusions

This forensic analysis of the CGS intake screens, based on hydraulic and biological literature, has identified three mechanisms that appear to result in extremely low entrainment of juvenile Chinook Salmon. These are, in order of apparent importance: (1) physical deflection of fish away from the screening material by the pressure and velocity fields at the T-screen structure’s nose cone (hydraulic bypass), (2) behavioral avoidance of the screen structure by drifting fish that are stimulated by the pressure and velocity gradients upstream of the screen structure’s nose cone, and (3) high ratio of sweeping flow to approach and through-screen velocities at the screen pores that prevents entrainment despite pore dimensions that would allow fish to enter.

The major conclusion is that the CGS intake’s near total lack of entrainment arises from the cylindrical screening structure, its placement in high-velocity river current, and the attendant hydraulic patterns around it. Hydraulic bypass alone may reduce potential entrainment. The low entrainment is likely aided by fish behavioral avoidance stimulated by those same hydraulic patterns. This conclusion differs from the historical research and regulatory attention given to the actual porous screens—their pore size, through-screen velocity, approach velocity, and sweeping flow. In CGS’s T-screen structures, the screen is hydraulically a minor part of the structure. Questions remain, however, over the effects, alone or interactive, of pore size, approach velocity, through-pore velocity, and sweeping flow on entrainment of the relatively few fish that enter the water layer close to the screen face. Since measurements of hydraulics and fish responses are problematic at such detailed scales, these questions may require CFD modeling paired with experimental observations of fish responses (such as by Stewart et al. 2014).

A likely protective scenario from the fish’s perspective (Coutant 1999) can be envisioned for juvenile salmon encountering the CGS cylindrical screen structures in the Columbia River:

• Downstream-migrating Chinook salmon and steelhead fry are temporarily in the river in spring, swimming gently head upstream into the current while in a drifting migration downstream. The gentle swimming generates a flow field around a fish’s body that is registered by its lateral line system and establishes the steady-state condition. That register is not static, but fluctuates within the bounds of change caused by the normally turbulent river flow.

• As a fish drifts downstream toward the cylindrical screen structure, it encounters a decrease in velocity and increase in pressure caused by the screen structure’s nose cone. The velocity and pressure changes sensed by the lateral line system are beyond those characterizing the baseline ambient condition.

• The fish is likely to react, which may be an instantaneous startle response causing the fish to deflect laterally and away from the structure.

• The fish may initiate more active swimming behavior, usually in the form of a short burst of swimming, to avoid the sharp pressure and velocity changes at the nose cone and bow wave.

• During or before these reactions, the fish enters the bow-wave-like diverging flow that moves the fish away from the nose cone and lateral to the face of the screen that forms the sides of the cylindrical T-screen structure.

• If the fish is on a trajectory to a side, top or bottom of the cylinder axis, it encounters the lateral velocity changes associated with the bow wave, and is deflected further away.

• The physical spreading of the flow around the cylinder in the form of the bow wave moves the fish around the cylinder. Inertia of the fish mass and avoidance swimming or orientation behavior counteract the pull of water into the screen’s orifices thus minimizing entrainment and impingement.

• While the fish is responding physically and behaviorally to the velocity changes associated with the bow wave, water is entering the screen through its own independent trajectory (flow net) as shown in model studies (Alam et al. 1974; NAI and ASA 2011a). For most of the screen area, the sweeping velocity greatly exceeds the approach and through-screen velocities so fish are not entrained.

• Any fish at risk of being impinged on the screen by in-flowing water (despite deflection by the bow wave) is rapidly washed off by the river’s strong sweeping velocity.

A cylindrical T-screen structure placed axially to ambient flow direction, in general, offers more opportunities for entrainment reduction than most screening systems reviewed by Turnpenny and O’Keeffe (2005) or published subsequently. Analysis of the CGS screening system differentiates between two different approaches to entrainment reduction: (1) by hydraulics and biological responses created by the structure holding the screens (nose cone of CGS’s cylindrical T-screen) and (2) hydraulics and biological responses at the fish-water separation surface such as traditional passive mesh screens, bar racks (Albayrak et al. 2020a, 2020b), trash racks (Raynal et al. 2013), curved bars (Beck et al. 2020), louvers (Albayrak et al. 2018), and flat-plate wedge-wire screens (de Bie et al. 2018) whether placed perpendicular to flow or angled. Cylindrical screens placed with the axis perpendicular to flow, such as tested by Weisberg, et al. (1987), are also in the second category. For most screening systems, screen mesh size is the main excluding feature according to Turnpenny and O’Keeffe (2005; sized to the shape and size of the target fish) although a sweeping velocity in angled screens can direct fish to a bypass.

Turnpenny and O’Keeffe (2005) missed an opportunity to make the distinction in entrainment reduction discussed above. They reviewed the design and performance of the ‘Passive Wedge-Wire Cylinder (PWWC) Tee-Screen’ similar to the CGS screen system but focused on the mesh attributes of through-screen velocity and slot width. Velocity of flow past the screen structure and screen siting were discussed only as these affected debris accumulation. They provided an example of replacement of an existing screen system with an array of passive wedgewire cylindrical screens (T-screens with 3 mm slot width) that reduced seasonal entrainment from thousands of lampreys to near zero (Frear and Axford 1991, as cited in Turnpenny and O’Keeffe 2005). They did not analyze why there was such remarkable reduction. Notably lacking was any consideration of hydraulic or behavioral effects of the screen structure.

This analysis of entrainment at the CGS differs from the typical forensic analysis by its use of surrogate data. Although the monitoring data that demonstrates near complete lack of entrainment was obtained from the CGS screen system, the hydraulic and biological evidence is derived from published literature. Thus, the conclusions are most properly viewed as a hypothesis. Both the hydraulic and biological responses are testable on the CGS screen system (assuming safety issues can be overcome) or on similar T-screens in flowing water.

Application to screening technologies generally

This forensic analysis offers several insights that may be useful for design and deployment of screening technologies in general.

• Foremost, it may shift attention more toward design and placement of the screening structure than to screen performance, per se. The way an entrainable organism fits through the screen’s holes or is impinged on the screen surface, while important, can be overshadowed by features of the structure that holds the screens.

• The greatest entrainment-reduction effect of the T-screen structure appears to be through its surrounding hydraulics, in particular the bow-wave-like hydraulics caused by the nose cone that affects direct hydraulic bypass and behavioral avoidance. Whether the screening structure is a T-screen or other device, attention to the hydraulics upstream of the actual screen may enhance overall entrainment reduction.

• Optimizing hydraulic bypass likely depends on understanding influences of nose cone shape, particle (fish) inertia, river speed, and intake suction strength that likely would determine how well particles diverge outwardly from stream lines just downstream of the nose cone.

• Focus on pressure and velocity changes upstream of the T-screen’s nose cone that stimulate avoidance behavior is a different sort of behavioral guidance than the behavioral technologies reviewed by Turnpenny and O’Keeffe (2005; bubble curtains, electric barriers, acoustic guidance, lights, velocity caps, turbulent attraction).

As an example, these insights would have been useful for the design and placement of T-screens as an option for entrainment-reduction technologies for cooling-water intakes in the USA. Analyses of alternative technologies were mandated by the U.S. Environmental Protection Agency under Section 316(b) of the Clean Water Act. I served as a peer reviewer of reports for 16 power stations and other industrial water users from 2017 into 2020. Cylindrical wedge-wire screen arrays were often proposed as an alternative to present once-through cooling or a closed-cycle recirculating system (cooling towers). Multiple, large T-screen units (20 to over 100 depending on slot width) were schematically arrayed in blocks in existing intake canals where there would be little or no axial flow to the units. The number of units needed was determined by the required water withdrawal and the screen criteria for through-screen and approach velocities for example slot sizes. None were schematically located in the main river current where there would be noticeable ambient velocity and only one array was placed single file extending from a channel-side bulkhead. Had the schematics placed screen units in river flow to take advantage of the major entrainment reductions seen in the CGS screen units, the number of screen units (and cost) would have been reduced markedly and the alternative viewed more favorably.

Deployment of screening technologies in many countries requires governmental regulatory approval, so it seems appropriate for the hydraulic factors providing entrainment reduction by a riverine T-screen structure like those at CGS be included in facility design criteria. These factors may be more important than traditional criteria at the screen face. With such criteria, permitting authorities could foster the testing and deployment of more effective screening systems.

What those additional criteria should be is ripe for further research. One criterion could be inclusion of structural components upstream of the actual screens that provide hydraulic bypass of the screen. An ambient velocity criterion may be useful for formation of an effective bow wave. Velocity-change or pressure-change criteria upstream of a screen structure might be useful for stimulating behavioral avoidance. These factors appear inherent in the axial-flow T-screen at CGS and may be feasible also for other screening systems. Each factor could be optimized to enhance the effect, tailored to the requirements of a site. The options for structures that modify hydraulics ahead of screens and deflect otherwise entrainable aquatic life seem endless.

Geolocation

Benton County, Washington, USA. 46.473500, −119.2646525

Acknowledgements

Benjamin Mater, Wayde Whitehead, the journal’s two anonymous reviewers and the Associate Editor provided valuable comments on early versions. I thank Stonybrook University, document custodian, for permission to cite the studies conducted for the Entergy Indian Point Energy Center (IPEC), and the principal investigators for those studies, Mark Mattson (Normandeau) and John Young (ASAAC) and their staffs for the valuable flume studies.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

Funding was provided by Energy Northwest for the initial analysis (Contract No. 00339040; Shannon Khounnala, Project Manager).