Sodium chloride as a tracer for hydrodynamic characterization of a shallow maturation pond

Hydraulically ponds are conceptually simple, in principle. Design criteria must lead to adequate hydraulic retention time (HRT) for the required biological reactions. However, the HRT value is only theoretical and can hide short-circuiting, dead zones (i.e., isolated or inaccessible regions of the pond), stagnant areas (regions where trapped fluids react weakly with the active volume), dispersion degree, thermal stratification, wind influence, etc, resulting in an actual HRT that differs from the theoretical and, possibly, a reduction in treatment efficiency. In polishing or maturation ponds, designed primarily to remove pathogens, the influence of the actual HRT on efficiency is even greater. Therefore, knowledge of the real hydraulic behavior is critical for their proper operation.

In most research on ponds, the hydraulic factors are determined via the stimulus-response technique, using tracers. Levenspiel (2000) presents one of the best-known theoretical references on this subject. The tracers available include radioactive, fluorescent, dye, saline and even biological (using bacteria) types. Each type has advantages and disadvantages. Those relating to natural wastewater treatment systems can be found in, for instance, Chazarenc et al. (2003), Camargo Valero & Mara (2009), Alvarado et al. (2011), Lange et al. (2011) and Matos et al. (2015). Saline tracers, for example, are usually cheap, and easy to acquire and quantify (with a simple conductivity meter), are non-toxic, and present no risk of environmental pollution or human health consequences. They can, however, bring disadvantages such as adsorption (by organic matter, biomass or sludge in ponds), high solution density and the need to use large quantities because of the high background concentrations usually found in sewage. Assessment of some of these practical issues, as related to a simple salt (sodium chloride), was one aim of this study.

This paper comprises the main results from saline tracer tests in a shallow maturation pond designed for a small population. The tracer used was common table salt (sodium chloride) in aqueous solution.

The experiments were undertaken at the Center for Research and Training in Sanitation UFMG-Copasa (CePTS), at the Arrudas wastewater treatment plant (WWTP), which receives municipal sewage from Belo Horizonte, Brazil (latitude 19°53′42″ S and longitude 43°52′42″ W). This has a Cwa (humid subtropical) climate according to the Köppen classification, with a mean annual temperature of 22.1 °C and mean annual rainfall of 1,540 mm. After preliminary treatment (coarse and medium screening followed by grit removal), a fraction of the wastewater goes to the experimental treatment plant.

The experimental apparatus consisted of a shallow maturation pond, designed for post-treatment of the effluent from a UASB (Upflow Anaerobic Sludge Blanket) reactor. A second shallow pond in series, not studied here, has longitudinal baffles. The system was designed to serve a population equivalent of 250 and receives approximately 40 m³/d of inflow (design value). The pond studied is 25.80 m long and 6.05 m wide (at half depth). Its mean depth is 0.80 m with an average sludge height of 0.34 m (bathymetry by Possmoser-Nascimento et al. (2014)). The pond's length/width ratio is 4.3:1, the inner dyke slope is 1:1 and the theoretical HRT is 3.1 d (or 2.1 d excluding the volume occupied by the sludge).

The tracer was common table salt (sodium chloride) in aqueous solution. The tests were performed using the stimulus-response technique (Levenspiel 2000) by means of tracer pulse injection at the pond's inlet and measurement of electrical conductivity at the outlet. The conductivity readings were taken with Global Water WQ301A-O probes (range 0 to 5,000 μS/cm), coupled to a GL500-7-2 Global Water^® data-logger and YSI 600XLM V2^® probes (range 0 to 100 mS/cm) with internal data-loggers. The sensors were attached near the outlet, 0.10 m below the liquid surface (it was not possible to place them outside the pond). Measurement frequency varied with the data collection schedule, the device storage capacity and the test phase (beginning or end), comprising intervals of 10 minutes. The probes set the conductivity automatically to a reference temperature (25 °C).

The tracer solutions were prepared in 200 and 800 L tanks (household water reservoirs), with salt addition and continuous homogenization. The pond's natural conductivity was measured prior to salt addition to define the background values. The amount of salt required was established by exploratory testing, to determine the minimum quantity (by volume) that could be measured against the background and peak levels below the probe maxima. In order to reduce the tracer volume, to approximate the application to a pulse-type injection and to facilitate liquid mixing, the lowest water volume possible was used for dilution, taking into account NaCl's solubility limit at ambient temperature. The tracer solution was introduced into the pond slowly in order to reduce semi-phase separation of the (denser) saline solution, but always in less than 2% of the pond's theoretical HRT. This was done to avoid mischaracterizing the pulse injection, following recommendations by Bracho et al. (2009). The flow in the ponds was also assessed before and during each test. Conductivity values (μS/cm) were converted to tracer concentration (mg/L) using the calibration equation presented by Possmoser-Nascimento (2014) for the same system ([tracer] = 0.5258*(conductivity) − 3.4836). Conditions for the four tests are summarized in Table 1.

It was possible to determine the mean actual HRT from the tests results as well as the variances, as the tracer concentration curve from the field tests can be defined by a discrete distribution of time intervals and tracer concentrations (Levenspiel 2000). It was also possible to obtain dispersion numbers and other hydraulic indexes of mixing and short-circuiting (Levenspiel 2000; Metcalf & Eddy 2003; Kadlec & Wallace 2009).

Figure 1 shows the tracer concentration measured in the effluent (mg/L) over time (days) in each test. Zero on the x-axis corresponds to the start of tracer injection. Tests were conducted over 17 to 35 days, considerably exceeding, in all cases, the general recommendation in the literature that tracer tests should be performed for a minimum of 3 times the theoretical HRT. According to Headley & Kadlec (2007), tracer responses from numerous experiments indicate that impulse-type tracer tests are typically complete after 4 times the theoretical HRT. In this study, the tracer was detected in the outlet throughout the entire tests.

The shapes of the curves in the first three tests were similar. In general, they indicate high dispersion and a tendency to complete mixing, with quick arrival of the tracer at the outlet characterizing short-circuiting in all tests. The occurrence of short-circuiting was confirmed by experimental tests with drogues and dyes, and simulation using computational fluid dynamics (CFD). These points are outside the scope of this paper, because of the complexity involved in their description and interpretation, but are detailed in Passos (2017).

It appears that a new tracer plume appeared approximately 10 days after the beginning of each test. This could be explained by the hydrodynamic behavior of the pond, which might have allowed the emergence of fluid streams passing through the outlet and returning later due to internal recirculation. Another hypothesis would be the subsequent slow release of saline tracer that could have remained in the pond in stagnant zones or adsorbed by the sludge. Anyway, the duration of the tests (far more than 3 times the theoretical HRT) and the slow output of the tracer, with the possibility of subsequent appearance of another portion of the fluid, suggested that tests in similar conditions to this study should be performed using more than 10 times the theoretical HRT to obtain additional information about the hydrodynamic behavior of the unit under study.

The oscillations in all curves showed daily cycles. These are thought to arise from thermal stratification, causing movement of part of the tracer solution to the bottom, followed by destratification with vertical mixing. This hypothesis is supported by measurements of the vertical profiles of temperature, dissolved oxygen (DO), pH and oxidation-reduction potential (redox). Typical examples of such behavior are presented in Figure 2, which shows the temperature, DO, pH and redox levels for more than two days. The triangular markers and dashed lines in the figure represent measurements taken near the pond surface, and the dotted markers and solid lines represent those from the bottom of the pond, with the x-axis representing the time of day and the vertical lines separating the days.

During this test, thermal stratification was observed from around 07.00 hrs, with destratification starting around 22.00. In other words, vertical mixing occurred in the pond for only for about 9 hours. DO levels on the surface were very high, as supersaturation took place during most of the day, because of hours of sunshine and the high photosynthetic activity in the shallow pond. Naturally, pH values were also higher on the surface.

DO was largely absent at the bottom of the pond and pH values were lower than at the surface. The data reveal a sharp rise in DO concentration and pH at the bottom of the pond after temperature equalization (destratification), validated by positive redox values at that time. This is probably because the liquid from the pond surface, with its higher DO concentrations and pH, was dragged down to the bottom – i.e., mixing occurred. Thus, DO and pH levels were equalized throughout the vertical profile, with vertical mixing occurring about an hour after temperatures had equalized. The same behavior was observed in the other tests. Destratification is mainly attributed to the cooling of the surface layers (related to air temperature cooling), so that the surface layers become denser, causing mixing.

The bottom probe in the pond was located so that it was very close to the sludge interface but not immersed in it. It thus represented the bottom of the liquid in the pond, not the pond itself.

The results of CFD modeling of the pond, including its thermal balance, confirm this behavior (Passos 2017).

An important aspect of the daily stratification and mixing cycles is that the saline tracer was likely to be involved. This would be true even if some of it stayed on the pond bottom due to its higher density, thus causing the concentration oscillations shown in Figure 1. Thus, Figure 2 suggests that the tracer sensor (at the surface) probably recorded more and less concentrated fluid layers through the cycle of stratification and vertical mixing. This would certainly influence the shapes of the curves and, consequently, the hydraulic parameters that could be extracted from them. No records of pond tracer tests were found in the literature showing oscillations like these, perhaps because of the measurement frequencies in other pond studies or their thermal profiles. Readings were taken every 10 minutes in this study and this might be why the oscillations were recorded.

It is believed that this is a novel observation. It is thought to be a function of the pond's shallow depth, and the relationship between the air and liquid temperatures, arising from its tropical location. Other tracers, which would be completely and uniformly dissolved throughout the liquid column, might not exhibit this behavior, which is associated with the saline tracer's density variations.

During the tests it was observed that an apparently new tracer plume appeared approximately 10 days after the start. This could be explained by the pond's hydrodynamic behavior, allowing the emergence of fluid streams passing close to the outlet but returning later due to internal recirculation. Alternatively, the effect might be produced by the slow release of saline tracer either held in the pond in stagnant zones or adsorbed in the sludge. Dye tracer tests – the next step in this research – will hopefully clarify the issue but the CFD hydrodynamic studies (Passos 2017) demonstrate recirculation flow lines along the entire length of the pond.

The mean HRTs were higher than predicted in theory, and, in most cases, more than double. Hydraulic parameters such as mixing and short-circuit indexes, volumetric efficiency, dead zone fractions and hydraulic efficiency (Persson et al. 1999; Levenspiel 2000; Metcalf & Eddy 2003; Kadlec & Wallace 2009) also presented unreliable values, differing substantially from what was expected. Because of this, these values are not discussed here.

According to Levenspiel (2000), the mass balance of tracers in reactors does not permit any retardation of residence time distribution (RTD) curves (which may result in real HRTs exceeding theoretical values), so that potential explanations include incorrect flow rate measurements, tracer adsorption and/or non-compliance with the closed-vessel condition. All of these factors are possible in this case and can be checked using another tracer, but the most likely factor seems to be the solution density. The high concentrations of the saline tracer, required to yield a significant concentration spike above background wastewater concentrations, resulted in substantial density effects, with the heavy tracer impulse sinking preferentially to the bottom and providing unrepresentative results. Headley & Kadlec (2007), in their practical guide to studies of tracers in wetlands, draw attention to this fact, reporting that the process was confirmed by Chazarenc et al. (2003) and Schmid et al. (2004).

The maximum tracer solution concentration in the tests was 300 kg/m³, sufficiently below NaCl's solubility in water at 20 °C (360 kg/m³) to allow dissolution and mixing in the preparation tank. Even for a small treatment system like that discussed here, the amounts used of salt are very high (around 300 kg for a 125 m³ pond). This can make transport difficult for standard vehicles, especially as WWTPs are often far from city centers and commercial zones. The large amount of salt required also makes this tracer rather impractical for use in work with larger ponds.

Due to the high salt concentration expected at the pond inlet zone, it would be wise to consider whether this might inhibit the biomass. No specific analyses were carried out in this respect for this study. It is noted, however, that the high initial salt concentration was dispersed relatively quickly in a larger volume of water in the pond, thus lowering it. No visible changes, such as algal death or color changes, were detected. Equally, there was no reduction in removal efficiency during the physical, chemical and microbiological monitoring of the pond's performance in the same period. This is an important aspect to consider in relation to the use of sodium chloride as a tracer, however.

Homogenization of the solution required substantial effort. Preparation for each test required about six hours of effort by two people. Manual mixing of the solution is very difficult. The most practical method, when mixing the solution manually, is to half fill the tank with water, adding most of the salt while mixing until saturation is achieved, then adding sequential amounts of salt and water, mixing further until solution saturation is achieved each time. It is estimated that the material for each test cost US$100, which is cheap compared to other tracers associated with more specific equipment.

It is easy to measure the effluent salt concentration using an electrical conductivity sensor, which can provide semi-continuous measurements during the test periods. There is no need to collect samples and analyze them in a laboratory, as can be the case with other tracers.

Density is an important consideration, since the sodium chloride tracer, which is dense, tends to sink to the bottom of the pond and travel towards the outlet without mixing with the fluid above. The specific gravity of the solution entering the pond was estimated at about 1.04. When working in wetlands, it is recommended that the tracer density should be within 1% of that of the wastewater environment (Headley & Kadlec 2007). In the pond used in this study destratification occurred daily, favoring vertical mixing and leading to smaller tracer peak values every day. In deeper ponds, salt tracer tests in more stable thermal stratification scenarios could be affected significantly by the solution's density, possibly inducing artificial short-circuiting.

It is recommended, for similar studies in ponds, that:

• Careful consideration is given to the possibility of adjusting the tracer solution density before starting a salt tracer test (by mixing with lower density substances, and controlling the solution's concentration and temperature)

• Tests are done, preferably, in periods when vertical mixing in the pond is observed, in the case of long-lasting mixing periods.

• Instantaneous addition of the whole volume of salt solution to the pond is avoided.

However, despite all the care that can be taken, it is emphasized that the tracer quantities, the solution volumes required, and the resulting application rates (to comply with the pulse injection concept) and density issues, are factors with high impact and interference in the pond's hydrodynamic conditions, which are generally not acceptable for tracers. It can then be considered that table salt is not a good tracer for such systems.

The shallow maturation pond showed a tendency to mix completely. Tracer peak anticipation was also very pronounced. These results may indicate the occurrence of short-circuiting resulting from the hydrodynamic behavior of the pond, thermal stratification in the vertical column or even a combination of these factors. Time-series effluent concentration values clearly indicate daily stratification and mixing cycles, which were corroborated by vertical measurements of temperature, pH, DO and redox in the pond's liquid column. The mean HRTs were higher than the theoretical ones, and mixing and short-circuiting index values were also unreliable. It is probable that another, completely dissolved, tracer, which would be present uniformly throughout the liquid column, would not show this behavior, which is associated with the variable density of the saline tracer. The slow output of the tracer, with the possibility of subsequent appearance of another portion of the fluid, suggested that tests in similar conditions to this study should be performed using more than 10 times the theoretical HRT.

The results of this study have enabled practical consideration of the use of NaCl as a tracer in ponds. NaCl is readily accessible and easy to quantify using electrical conductivity, but large quantities are required, which may make transport difficult.

Solution preparation and homogeneity control demand substantial effort and time. In large ponds, application is almost impossible because of the amount of salt needed. Solution density may also result in unrealistic responses in ponds, since it is difficult to avoid tracer concentration gradients along the vertical profile, especially in (deeper) ponds with longer thermal stratification periods. In general, therefore, sodium chloride is not a good tracer for such systems. It can be used successfully in systems with low HRTs and with porous media units (constructed wetlands, for example), since the effects of density are reduced and the amount of solute required is low.

The authors would like to thank COPASA and the Brazilian agencies CAPES, CNPq and FAPEMIG for their support to the research. This research was also part of an international program financed by the Bill & Melinda Gates Foundation for the project ‘‘Stimulating local innovation on sanitation for the urban poor in Sub-Saharan Africa and South-East Asia’’ under the coordination of Unesco-IHE, Institute for Water Education, Delft, the Netherlands.