5 A turbulence chamber appropriate for laboratory studies of scalar fluxes at the sediment-water 6 interface was developed and its turbulence characterized using particle image velocimetry (PIV). 7 The chamber is capable of reproducing a wide range of turbulence levels, including those typi-8 cal of low energy environments such as lake and reservoir bottom boundary layers (BBL), which 9 motivated the present effort and against which the chamber's turbulence levels were calibrated. 10 The

more »

... turbulence is forced by three peristaltic pumps plumbed to six equispaced orifices 11 on the top cap of the chamber, issuing either as momentum sources (jets) in the vertical direction 12 or momentum sinks (potentials). The pumps operate continuously but randomly change direction 13 to generate a horizontally homogeneous turbulent region near the sediment-water interface due to 14 jet interaction. The PIV measured turbulence intensities show a quadratic relationship with pump 15 speed (RPM), with typical environmental BBL turbulence intensities seen at low pump speeds 16 (∼10 RPM where the maximum speed is 100 RPM). The chamber design can easily be adapted 17 to the physical constraints of the environmental system under study, such as the work with con-18 taminated sediments that motivated its development. Relative to similar facilities, the developed 19 chamber represents the first facility with well characterized turbulence calibrated to environmental 20 turbulence levels. 24 Scalar flux at the sediment-water interface and sediment scalar demand (e.g., oxygen or nitrate) 25 are important quantities in lake and reservoir management. Sediment oxygen demand (SOD) is a 26 common metric in bottom boundary layer (BBL) chemistry, with high SOD often leading to low 27 dissolved oxygen (DO) levels and anoxic conditions. Low DO can cause fish kills, bacterial com-28 munity changes and the production of undesirable compounds (e.g., hydrogen sulfide, ammonia, 29 orthophosphate) (Beutel et al. 2007) and increased activity for heavy metals such as mercury 30 (Mackenthun and Stefan 1998). 31 Studies typically examine oxygen flux at the sediment-water interface by making finely re-32 solved measurements of (i.e., profiling) the scalar concentration boundary layer (CBL, also called 33 the diffusive boundary layer, or DBL). The CBL is a region, millimeters thick, where scalar con-34 centration transitions from the bulk water concentration (C ∞ ) to the concentration in the sediment 35 surface layer, and hence it sets the scalar flux resulting from the mean concentration gradient. Tur-36 bulent stirring will directly affect δ CBL , modifying the CBL thickness and hence the strength of 37 the driving mean gradient (Lorke et al. 2003). 38 Hondzo (1998) performed experiments in a recirculating open channel flume and showed how 39 changes in the friction velocity u * (i.e., changes in mean flow strength) influence oxygen fluxes at 40 the sediment-water interface. Using a 2-D laser Doppler velocimeter (LDV), he made measure-41 ments of the Reynolds shear stress u w to estimate u * , finding a linear dependence of the DO 42 transfer rate on u * and δ CBL proportional to u −1 * . 43 Lorke et al. (2003) used high resolution Doppler current profilers to measure turbulence in the 44 momentum boundary layer and microelectrodes to measure DO in a lacustrine BBL. They found 45 increased turbulence levels as parameterized by turbulent dissipation, , led to decreased δ CBL . 46 The Batchelor length (L B = ( νD 2 ) 1/4 was identified as a direct estimate of δ CBL . Using scaling 47 arguments for a turbulent boundary layer, Lorke et al. (2003) showed that the scaling δ CBL ∝ 48 ν u * Sc −1/2 , where Sc is the Schmidt number defined as ν D , proposed by Hondzo (1998), is equivalent 49 2 Rusello, July 31, 2012 to using the Batchelor length scaling for δ CBL . Thus in systems where the parameterization of 50 turbulence by u * is appropriate, e.g. energetic systems such as rivers, the two scalings should 51 provide similar values. In lower energy systems where u * is not an appropriate scale for turbulent 52 velocities, such as low energy systems like the BBL of lakes and reservoirs, the use of L B is more 53 appropriate (Lorke et al. 2003). 54 Horizontal advection also affects δ CBL . Lorke et al. (2003) noted non-stationary bulk con-55 centrations (C inf ) had a large influence on CBL thickness, estimating a 10% uncertainty in C inf 56 resulted in 30% uncertainty in the thickness of the CBL. While C inf is important in determining 57 the CBL thickness, it is ultimately turbulence near the sediment-water interface which physically 58 controls δ CBL at a given bulk concentration. 59 Present Need and Chamber Design 60 Onondaga Lake is a medium sized, urban lake located northwest of downtown Syracuse, NY 61 and referred to as the most polluted lake in North America (Effler 1996; Effler and Hennigan 62 1996). The production of soda ash in Solvay on the south shore of the lake resulted in large 63 amounts of ionic pollution in the early 20th century, while the waste water effluent from Syracuse 64 and surrounding communities introduced high nutrient loads over several decades. The pollutant 65 of primary concern, however, is mercury. 66 Starting in 1946 and continuing through 1972, approximately 70 metric tons of mercury were 67 dumped into the lake as a byproduct of chlorine gas production (Effler and Hennigan 1996). This 68 has resulted in the entire lake bed being declared a Superfund site. Because of high nutrient 69 loading and stratification, anoxic conditions persist at the sediment-water interface for much of 70 the stratified season. While the low DO levels are ecologically undesirable, oxidation-reduction 71 reactions involving sulfur, which occur in the low DO conditions, create a preferential pathway for 72 mercury compounds to move out of the sediment and enter the water column. Recent remediation 73 work has focused on addressing the low DO levels through electron acceptor augmentation (Auer 74 et al. 2010). 75 SOD studies related to the remediation work require a facility which is able to handle mer-76 3 Rusello, July 31, 2012 cury contaminated sediments, produce turbulence similar to BBL conditions, and allow changes 77 in turbulence caused by various remediation methods to be simulated. Auer et al. (2010) dis-78 cusses proposed remediation methods which include an oxygen bubbler system and a nitrate slurry 79 injection, expected to enhance and suppress ambient levels of turbulence, respectively. 80 Researchers such as Hondzo (1998) and O'Connor and Hondzo (2008) worked in open channel 81 flumes, which replicate boundary layer flow observed in streams and rivers. However, in lake BBLs 82 and other low energy systems, the canonical turbulent boundary layer is potentially a poor model 83 for flow structure, and u * a poor parameterization of turbulence in these systems (see §5 and Lorke 84 et al. (2003)). 85 In addition to the physical considerations of producing a suitable low energy environment to 86 simulate Onondaga Lake conditions, the practical aspects of safely working with mercury contam-87 inated sediments further constrain experimental work. While an open channel flume could produce 88 low energy conditions, the large volume of water and de-contamination concerns eliminate it as a 89 practical facility to study SOD and work with contaminated sediments. A facility producing tur-90 bulence levels of the same magnitude as in the Onondaga BBL, and in particular low turbulence 91 levels where L B is a reasonable estimate of δ CBL , would be suitable for studying SOD in the lab-92 oratory. To meet both the physical and practical aspects of studying SOD in Onondaga Lake, a 93 small, low cost, turbulence chamber with these characteristics was developed. 94 Chamber Design 95 Beutel et al. (2007) developed a laboratory SOD measurement chamber forced by a recircu-96 lating peristaltic pump. A small jet (orifice diameter unreported) was introduced 2-3 cm above 97 the sediment surface with the jet axis parallel to the surface. Measurements were made under 98 quiescent, moderately (defined as jet velocities of 3-4 cm s −1 ), and highly mixed (defined as jet 99 velocities of 6-8 cm s −1 ) conditions. The authors acknowledge that their facility is sub-optimal 100 for the study due to horizontal spatial gradients in the mean flow. Results indicate an increase in 101 SOD as mixing level increased, but no quantitative turbulence or mean velocity measurements are 102 reported to allow a relationship between the two to be developed. 103 157 orifice spacing, would increase the merging of individual jets and create more homogeneous mean 388 and turbulent flows. 389 The behavior of the chamber at lower pump speeds was investigated by first considering the jet 390 Reynolds numbers, Re j , where U jet was determined from volume flow rate measurements and d is 391 15 Rusello, July 31, 2012 the inner diameter of the jet orifice, set by the tubing used to plumb the chamber. Re j estimates are 392 280, 680 and 960 for pump speeds of 50, 75 and 100 RPM, which for the oscillatory forcing of a 393 peristaltic pump are expected to be sufficiently large to support a turbulent jet. This was confirmed 394 by flow visualization (Figure 14) of a single jet entering the chamber which revealed strong stirring 395 indicative of turbulent flow for pump speeds at 25 RPM and greater. Below 25 RPM, the pulsed 396 forcing of the peristaltic pump increased stirring, but truly turbulent flow does not develop. This 397 suggests that jet-jet and/or jet-bottom boundary interaction leads to the development of turbulence 398 at pump speeds of 6 and 10 RPM, which was demonstrated to exist based on the velocity spectra. 399 Though conceptually similar to the random jet stirred tank of Variano and Cowen (2008) and 400 the jet stirred tank of Webster et al. (2004), there are key differences between the present chamber 401 and these two systems. These differences affect both the flow and operation of the chamber. The 402 always on operation of the peristaltic pumps means a jet orifice is an alternating source and sink 403 of momentum rather than just a source. While there are six orifices, there are only three pumps 404 used for the developed chamber, reducing the degrees of freedom from 6 to nominally 3. The 405 physically separated inlet and outlet and the reduced degrees of freedom create sources and sinks 406 of momentum associated with each pump that are not co-located, generating a net mass transport 407 from inlet to outlet within the chamber, which is zero only when averaged over sufficient time or a 408 spatial region containing a single pump's inlet and outlet. 409 The performance of the low RPM operation of the chamber is important given the measured in-410 situ turbulence levels. At low RPM, despite generating turbulence similar to measured lacustrine 411 BBL conditions, performance can be improved by increasing the jet spacing to water column 412 height ration and tuning the mean time between direction changes to maximize turbulence and 413 minimize mean flow. An early test of the chamber using a slightly larger diameter examined the 414 dependence of turbulence intensity on mean time between direction changes. Results showed a 415 linear increase in turbulence intensity and a linear decrease in mean flow as the mean time between 416 direction changes increased. We attribute this to the mechanical aspects of the peristaltic pumps 417 used to drive flow and the finite amount of time needed for them to change rotation direction. 418 16 Rusello, July 31, 2012 For continuity with ongoing SOD experiments, the mean time between direction changes was not 419 altered for the calibration and testing presented here, remaining fixed at three seconds. 420 Increasing the water column height or decreasing the jet orifice spacing will allow the jets to 421 merge more fully. Based on the results of Variano and Cowen (2008) , water column height would 422 need to be 300 mm or the orifice spacing would need to be 17 mm (requiring more orifices to 423 achieve uniform spacing) to obtain the water column depth to jet orifice spacing ratio of 6 needed 424 to ensure individual jets fully merge. Increasing chamber height has the added benefit of shifting 425 the calibration curve so a higher pump speed will be needed to achieve a given turbulence intensity, 426 potentially creating a more homogeneous velocity field and more fully developed turbulent flow at 427 lower turbulence intensities. 428 CONCLUSIONS 429 In-situ and laboratory measurements have shown the importance of turbulence in establishing 430 the rate of scalar fluxes, Many engineering applications require laboratory facilities to measure 431 SOD under controlled, repeatable conditions representative of the turbulent benthic environment. 432 Facilities for this type of testing have been hampered by a lack of quantitative measurements of 433 turbulence and mean flow levels as well as a lack of comparison to environmental turbulence levels 434 and mean flows. Often these facilities provide only qualitative assessments of mixing as moderate 435 or strong while a few researchers have made efforts to quantify the bulk turbulence in their facilities 436 using u * but in flows with strong spatial gradients. 437 The shear velocity, u * , used by many researchers to characterize turbulent fluxes, is not a 438 reliable representation of turbulence in the absence of boundary layer structure typically seen in 439 low energy lacustrine environments or where turbulence sources exist within the watercolumn, 440 such as internal waves, water column shear, or breaking surface waves in shallower environments. 441 Using vertical turbulence intensity, w 2 , as a comparison metric for laboratory and field studies 442 captures relevant physical information available in u * directly, namely the turbulence intensity as 443 well as the fundamental driving component of the turbulent vertical scalar flux, w c where c is the 444 scalar concentration fluctuation. It also permits the estimation of both dissipation and the Batchelor 445 17 Rusello, July 31, 2012 length scale for use in the direct scaling of δ CBL with dissipation utilized by Lorke et al. (2003) 446 without any assumptions on mean flow structure. Finally, w 2 is the most straightforward, robust, 447 and economical quantity to measure in lake BBLs. 448 A laboratory facility using a random array of jets driven by peristaltic pumps is used to produce 449 turbulence levels typical of a lacustrine BBL. The facility is characterized using values over a 450 homogeneous region from 10-20 mm above the bottom. Mean and turbulent flows are shown to be 451 essentially homogeneous in this region, and there is a predictable variation in turbulence intensity 452 and dissipation with pump speed. A calibration curve relating turbulence intensity to pump speed 453 for SOD testing was produced, allowing researchers the flexibility and control to develop SOD 454 versus turbulence intensity curves. The chamber design is easily modified for given physical or 455 chemical constraints and cheaply produced as a new facility given its minimal material content. 456 The use of peristaltic pumps commonly found in laboratories allows multiple chambers to be run 457 off of a single set of peristaltic pump drive units, facilitating replicate experiments as well. 458