Function of soil-based on-site wastewater treatment systems - Biological and chemical treatment capacity

(1)

IN

DEGREE PROJECT ENVIRONMENTAL ENGINEERING, SECOND CYCLE, 30 CREDITS

,

STOCKHOLM SWEDEN 2017

Function of soil-based on-site

wastewater treatment systems -

Biological and chemical treatment

capacity

LINNÉA TJERNSTRÖM

KTH ROYAL INSTITUTE OF TECHNOLOGY

(2)

TRITA LWR Degree Project ISSN 1651-064X

(3)

Function of soil-based on-site

wastewater treatment systems

- Biological and chemical treatment capacity

LINNÉA TJERNSTRÖM

(4)

(5)

Abstract

On-site wastewater treatment systems are among the main Swedish anthropogenic sources of nutrients causing euthropication of the Baltic Sea. Among on-site

systems in Sweden almost half have septic tank treatment followed by a soil-based system, in which the wastewater is treated through soil filtration. In this study a soil based technique for on-site wastewater treatment is studied where wastewater is filtered through a sand filter. Composite samples of influent and effluent at two sand filters in the area of Stockholm are sampled to determine their chemical and biological function and to compare their treatment capacity to requirements. Parameters within the scope of the study are tot-P, NH4-N, DOC, pH, turbidity and dissolved oxygen. Biological function was considered to be good in both systems as nitrification was high and the effluent had sufficient levels of dissolved oxygen suggesting aerobic conditions. Prevailing aerobic conditions in the sand filters would also indicate good reduction of organic substances which is the case for DOC as it was reduced by above 85 % for one site and almost 70 % for the other site. The overall high reduction of organic micropollutants in the systems, reported in

(6)

(7)

S

_{ummary in Swedish}

Decentraliserade system för rening av avloppsvatten är bland de huvudsakliga svenska antropogena källorna till näringsämnen som bidrar till övergödning av Östersjön. Bland decentraliserade system i Sverige är nästan hälften system med slamavskiljare följt av ett markbaserat system i vilket avloppsvattnet renas genom infiltration i jord. I denna studie studeras en markbaserad teknik i vilken

avloppsvattnet filtreras genom sand, en så kallad markbädd. En fältundersökning gjordes där samlingsprov av ingående och utgående avloppsvatten togs på två markbäddar i Stockholmsområdet för att bestämma deras biologiska och kemiska reningsfunktion samt att jämföra avskiljningen av fosfor i systemen med

rekommendationer från HaV. Parametrar som inkluderats i studien är totalfosfor, ammonium-kväve, löst organiskt kol, pH, turbiditet och löst syre. Biologisk

funktion ansågs bra i båda markbäddarna eftersom nitrifikationen var hög och utgående vatten hade tillräckliga halter av löst syre vilket implicerar att

markbäddarna var väl syresatta. Rådande syrerika förhållanden i markbäddarna antyder också att organiskt material bryts ned avsevärt, vilket är fallet för löst kol som reducerades med mer än 85 % i en av markbäddarna och med nästan 70 % i den andra. Den höga reduktionen av organiska mikroföroreningar som påvisats i markbäddarna i en annan studie tyder också på att biologisk funktion med avseende på avsklijning av organiska substanser är bra. Kemisk funktion, med avseende på avskiljning av totalfosfor, var inte tillräcklig då ingen av

(8)

(9)

Acknowledgements

I would like to thank my supervisor Gunno Renman from the Department of

Sustainable Development, Environmental Sciences and Technology at KTH for help and support with this degree project. Also I want to thank the FORMAS project RedMic for letting me be a part of the project and Wen Zhang from the Department of Land and Water Technology at KTH as well as master student Anqi Li from KTH for help at the lab during the lab analysis.

Finally I would like to thank friends and family who supported me through the work, read my manuscript and were there for me when I doubted what I was doing.

(10)

(11)

List of abbreviations

BOD Biological oxygen demand (mg O2/L). A measure of amount of

biodegradable organic substances in the wastewater. It is the amount of oxygen dissolved in the wastewater that is used to degrade organic substances during a certain time period and temperature by

microorganisms. BOD7 has a time period of 7 days and BOD5 of 5 days

DO Dissolved oxygen (mg/L)

DOC Dissolved organic carbon (mg/L)

HaV Swedish agency for marine and water management (Havs- och

vattenmyndigheten)

HRT Hydraulic retention time. The average time it takes for water molecules to travel through a system

IC Inorganic carbon (mg/L)

K Hydraulic conductivity (m/s). A measure of a soils’ capacity to conduct water. It is different for saturated and unsaturated conditions

N Nitrogen. Measured as tot-N (mg/L)

P Phosphorus. Measured as tot-P (mg/L)

PE Person equivalent. Average amount of wastewater pollutants originating from one person during one day. In Sweden 1 PE it is defined as a BOD7 load of 70 g O2 /L/day

SEPA Swedish environmental protection agency

TC Total carbon (mg/L)

TE Treatment efficiency (%)

(14)

(15)

List of explanations of expressions and words

Anaerobic conditions Conditions with absent of free oxygen

Aerobic conditions Conditions with available free oxygen

Aerobic microorganisms Microorganisms which use oxygen to oxidise

substrates

Black-box study A study where treatment efficiency of i.e a

sand filter is estimated by sampling of influent and effluent and calculated from influent (ci) and effluent (ce) concentrations as 100 (ci - ce) / ci

Effective precipitation The share of actual precipitation which is not

transpired by plants or evaporated from the soil

Biofilm Microorganisms covering a surface

Heterotrophic microorganisms Microorganisms which degrade organic carbon in order to gain energy

Hydraulic load Volume of primary treated wastewater applied

to the sand filter surface per day (m3_/day)

Hydraulic surface load Primary treated wastewater applied to the

sand filter per m²and day (cm/day)

Mass balance study A study where treatment capacity of i.e a sand

(16)

(17)

INTRODUCTION | 1

1 Introduction

Human activities impact aquatic environments like streams, wetlands, lakes, rivers and seas, with negative risks to the health of humans and animals as a

consequence. One of the main surface water quality problems today is

eutrophication which is due to the anthropogenic release of nutrients (P and N) into the natural environment (Khan & Mohammad, 2014). Abundance of P and N in surface water results in excessive growth of aquatic vegetation and algae which can cause oxygen depletion on sea and lake floors (Smith, 2003). Also it causes blooms of the blue-green algae which is a health risk for animals and humans (Khan & Mohammad, 2014).

Globally, main human sources of P and N are agricultural drainage and wastewater from industry, municipal WWTP:s and on-site wastewater treatment systems (Khan & Mohammad, 2014). In Sweden there exist around 625 000 households connected to on-site wastewater treatment systems dimensioned for up to 200 PE (Olshammar et al., 2015) and these are contributing significantly to the P load to the Baltic sea, which is heavily affected by euthrophication (HaV, 2016). On-site systems for wastewater treatment are estimated to contribute with 15 % of total anthropogenic P load to the Baltic sea, with municipal wastewater treatment systems and industries contributing with about as much (18 and 19 % respectively). When it comes to N release to the Baltic sea on-site systems contribute with a 3.8 % share of total anthropogenic load (Ejhed et al., 2016).

In order to limit P and N discharges from on-site wastewater treatment systems in particular the Swedish agency for Marine and Water management (HaV) has general advises (HVMFS 2016:17) for systems for less than 26 PE1_{to have a}

minimum reduction of 70 % for P for areas with a so called normal protection level. For on-site systems in areas with a so called high protection level 90 % reduction of P and a minimum reduction of 50 % of N is recommended (Table 1). Protection level is determined by the municipality based on sensitivity of the area and

recipient for each individual case (HVMFS 2016:17).

(18)

2 | INTRODUCTION

Another more recent concern within the field of wastewater treatment is release of organic micropollutants to aquatic envionments. These pollutants pose a risk to receiving waters even at low concentrations (ng/L or μg/L) (Barbosa et al., 2016). For example pharmaceutical residuals in water bodies can have negative effects on fish or other water living organisms as they are persistent and might therefore accumulate in the environment, as well as within organisms, up to high

concentrations (Larsson and Lööf, 2014). Swedish on-site systems for wastewater treatment has recently been detected as a potential source of pharmaceutical residues to the environment (Woldegiorgis et al., 2007) as well as other organic micropollutants (e.g biocides, personal care products) (Ridderstolpe et al., 2009). Table 1: General advises for minimum reduction of P and N and what effluent concentration it

corresponds to in on-site wastewater treatment systems (< 26 PE) from the Swedish agency for Marine and Water management (HVMFS 2016:17).

Phosphorus (tot-P) Nitrogen (tot-N)

Reduction (%) Effluent2_(mg/L) _{Reduction (%)} _Effluent2_(mg/L)

Normal protection level 70 3 -

-High protection level 90 1 50 40

Among the treatment techniques for on-site wastewater treatment systems (up to 200 PE) 26 % only have septic tank treatment. Almost half of the on-site systems have septic tank treatment followed by a soil-based system, with around 14 % being small sand filters (< 26 PE), around 30 % small infiltration systems with

groundwater as a recipient, 2 % being small sand filters with P-filter and 2 % being large infiltration systems or sand filters for 25-200 PE (Olshammar et al., 2015). As soil-based systems are a common technology for on-site wastewater treatment in Sweden and thereby stands for significant discharge of nutrients and other pollutants to water environments it is important to investigate their treatment capacity and their performance.

In this report the results of a field study on two soil based systems for wastewater treatment (sand filters) located in the area of Stockholm is presented. Results included in this report are analysis results of physical parameters (including turbidity and electric conductivity) and macropollutants (including organic substances and nutrients) as well as on-site measurements of pH and DO. The wastewater samples from the field study were also sent to Umeå University and Swedish University of Agricultural Sciences (SLU) were they were analysed for several organic micropollutants within the scope of the FORMAS project RedMic (see Blum et al., 2017).

(19)

INTRODUCTION | 3

1.1 Aim and study questions

This study aims at reviewing the literature on treatment capacity of soil based systems as well as evaluate the function of two soil-based systems for on-site treatment of wastewater. The questions the study aims to answer are:

• What is the treatment capacity of soil based systems in the literature? • How is the mechanical, biological and chemical function of the soil based

systems in the field study?

• How are their treatment capacity when it comes to reduction of P in comparison from the general advises from HaV?

1.2 Scope and limitations

(20)

THEORETICAL BACKGROUND | 5

2 Theoretical background

2.1 Wastewater constituents

Wastewater consist of inorganic and organic substances which can be either

dissolved, suspended or colloids (Wiesmann et al., 2007). Inorganics in wastewater are suspended solids, such as sand and clay minerals, and dissolved substances such as nutrients (e.g P and N) and a small amount of heavy metals and other metals (Wiesmann et al., 2007). Suspended organics include microorganisms (pathogenic and non-pathogenic) as well as residuals from food. Dissolved organics include biodegradable and non-biodegradable substances, and organic colloids include grease and oil drops (emulsions) as well as solid particles (Wiesmann et al., 2007).

2.2 Common water and wastewater quality parameters BOD

BOD is a measure of the amount of oxygen (mg/L) used by microorganisms to decompose organic substances in the wastewater during a specific time period at a specific temperature at a specific pH (Wiesmann et al., 2007). Wastewater with high BODcannot be released in environment since it consumes a lot of oxygen when it is decomposed which may cause anaerobicity in receiving waters (Siegrist et al., 2000).

TOC and DOC

Total organic carbon (TOC) is the amount of carbon in organic substances in the water and includes both suspended, dissolved substances and colloids. To get DOC suspended organics and colloids has to be removed from the sample, for example by a filter (Wiesmann et al., 2007)

tot-N, NH4-N and tot-P

(21)

6 | THEORETICAL BACKGROUND

(NH4+) (NH4-N) and between 10 and 30 % is in organic form. (Siegrist et al., 2000) Phosphorus in wastewater is either in dissolved form or particulate form. 75 - 100 % of the total P (tot-P) in wastewater is inorganic orthophosphates (H2PO4-, HPO4 2-and PO43- ) (von Brömssen et al., 1985) and the remaining is polyphosphates or bound in organic material (von Brömssen et al., 1985; Siegrist et al., 2000). pH and dissolved oxygen

One common parameter is DO. The lower the temperature of the water the higher is the saturation concentration of oxygen. At temperatures 12 °C saturation

concentration of oxygen in water is 10.8 mg/L (Vesilind et al., 2013). The pH of water is also in important parameter since discharges of water with a high or low pH compared to the natural pH in a receiving water can have negative impacts on aquatic life (Vesilind et al., 2013).

Turbidity

Turbidity is a physical parameter caused by suspended and colloidal matter present in the water which make the water hazy. It is measured by the Nephelometric Turbidity Unit (NTU). (Vesilind et al., 2013). As turbidity is a measurement of the haziness of the water it does not directly measure the amount of suspended matter. Turbid water may have negative impacts on receiving waters as it affect

photosynthesis since less light reach down into the water. Also suspended matter can impact fish negatively since it gets caught in their gills (USEPA, 2012).

2.3 Representative wastewater sampling

During wastewater sampling it is important that the sample is a good

representation of the wastewater. There are two different sample types, namely grab sample and composite sample. A grab sample represents a snap-shot of the quality of the wastewater at the specific time it was sampled. According to SS 02 81 48 (SIS, 1981) a grab sample can represent the wastewater quality only if variation in concentrations in the wastewater is low over time. If wastewater quality varies over time as well as the flow then a flow proportional composite sample better represents the wastewater. A composite sample is composed of a number of

samples and represents the mean water quality during a period of time, it could for example be sampled during 24 hours or a week. Flow proportional sampling is carried out by having a sampling frequency proportional to the flow with constant sample volume or constant time between the samples, but the sample volume is proportional to the flow (SIS, 1981). Composite samples can also be time

(22)

intervals and pooled into one sample. This can be accomplished by an automatic sampler, for example Isco 6712 (Teledyne Isco, 2015) It is basically composed of a number of grab samples and could therefore only be a good representation of the wastewater if the flow is constant. During sampling and transport from the site samples should be stored at a cold and dark place according to SS 02 81 48 (SIS, 1981).

2.4 Sand filter for on-site domestic wastewater treatment

A conventional sand filter is a unit for treatment of wastewater in which wastewater is infiltrated at a constructed subsurface below ground and filtered through sand (SEPA, 2008). A principal overview of a sand filter with preceding treatment and additional components can be seen in figure 1. Prior infiltration into the sand filter wastewater is treated in a septic tank.

The primary treated wastewater from the septic tank is pumped, or fed by gravity to a distribution well in which the wastewater is divided into a number of application pipes. From the pipes it is distributed into a application layer with coarse material for further infiltration to a layer composed of sand. After infiltration the wastewater percolates vertically through the sand layer under unsaturated flow conditions (Ridderstolpe, 2009) (SEPA, 2008) being subject to biological, chemical and mechanical treatment processes.

(23)

The water is collected in pipes in a drainage layer and directed to an outlet from where the water is lead to further treatment, or to a surface water recipient. Treated water might also leak to the groundwater if the bottom of the system is not

watertight (SEPA, 2008). Septic tank treatment

A septic tank (figure 2) is a common technology for sludge separation in small decentralized wastewater treatment facilities used prior more extensive treatment (Palm et al., 2002). It is used to prevent suspended solids in the wastewater to reach the sand filter since they may cause clogging and malfunction (Siegrist et al., 2000). The tank is usually constructed with three cambers and as the wastewater flows through suspended solids are separated by either flotation or sedimentation. Variations in incoming flow are leveled off and the septic tank serves as storage for the sludge (Palm et al., 2002). Suspended solids are reduced by 70 % in a septic tank and some organic particles are digested anaerobically resulting in a reduction of BOD with 10-15 % (von Brömssen et al., 1985) Reduction of P and N is low (Palm et al., 2002). As an example a reduction of 15-20 % for tot-P and 10-20 % for tot-N is reported in SEPA (1991).

Effluent concentrations from septic tanks varies in the literature. An average value between 7.2-16.3 mg/L is reported for tot-P (Appendix) and an average value between 36-100 mg/L. (Appendix) Expected concentration of nutrients from

(24)

domestic wastewater are according to guidelines from HaV (HVMFS 2016:17) 12 mg/L for tot-P and 80 mg/L for tot-N (the numbers are based on templates on daily water usage per person and daily load of tot-N and tot-P per person). DOC concentration in wastewater after sludge separation was reported in Zhang & Renman (n.d) to be on average 35.5 mg/L. BOD5 in effluent from septic tank was on average 320 mg/L in Matamoros et al. (2009). BOD7 are reported between 165-218 mg/L in the literature (Appendix).

Mechanisms for removal and retention of organic substances and N in sand filters As wastewater infiltrates into the sand after start-up of a new sand filter a clogging zone is developed at the infiltration surface since suspended material get caught in the pore structure at the same time as microorganisms on sand particle surfaces grow in numbers creating a biofilm on sand particles (Ridderstolpe, 2009). Infiltration capacity in the sand is thereby drastically reduced over a period of months to years, ending up at a more stable but lower infiltration capacity compared to before (Ridderstolpe, 2009). The low infiltration capacity causes a temporarily pool of water in the application layer above the infiltration surface each time wastewater is applied (Ridderstolpe, 2009).

The wastewater applied to the sand filter infiltrates slowly into the clogging zone and anaerobic conditions prevail near the infiltration surface. Deeper down in the filter, where the pores are less clogged and thereby larger the flow becomes

unsaturated (von Brömssen et al., 1985; Siegrist et al., 2000). It is in this transition zone, between more clogged and less clogged pores, most of the biological

treatment takes place by aerobic microorganisms in the biofilm (Ridderstolpe, 2009). Heterothropic aerobic microorganism use organic substances as a source of energy and transform them into CO2, H2O and nutrients (Ridderstolpe, 2009). Aerobic bacteria (chemolitho-autotrophic bacteria) transforms ammonium (NH4+) into nitrite (NO2-) and subsequently to nitrate (NO3-) by a process called

nitrification (Wiesmann et al., 2007).

Organic substances and N are not only subject to degradation by aerobic

microorganisms in sand filters. Dissolved organic substances in wastewater can be removed by volatilization and suspended organic material can get caught in the pore structure and thereby be removed from the wastewater (Siegrist et al., 2000). Organically bound N becomes ammonium (NH4+) through a process called

(25)

Mechanisms of P retention in soils and sand filters

Sorption is an important mechanism for P retention in soils. Phosphate, which is an anion, can be adsorbed to and form surface complexes with Al(III)/Fe(III)-(hydr)oxides in a soil (Eveborn et al., 2012b; Eveborn et al., 2014). Surface complexation is a strong bond in which the adsorbed ion more or less becomes a part of the surface (Gustafsson et al., 2010). The charge on the (hydr)oxides is pH dependent (Gustafsson et al., 2010), more positive at low pH, resulting in

phosphate adsorption to (hydr)oxides being strongest at low pH as shown in Gérard (2016). Also sorption of P to clay minerals is a significant mechanisms governing P retention in soils. Sorption of P to clay minerals is also pH dependent, being highest at low pH (Gérard, 2016). Phosphorus can also be retained in soil by precipitation, which means that the phosphate ion in the soil water leaves this phase and becomes a part of the soil as a solid mineral phase.

Precipitation/dissolution of Al(II)/Fe(II) phosphates (Al-P / Fe-P) as well as Ca phosphates (Ca-P) are examples of chemical processes of importance for P in a soil-water environment (von Brömssen et al., 1985; Eveborn et al., 2012b; Eveborn et al., 2014). These reactions strives for an equilibrium between the solid phase and the soil water phase. Precipitation of the dissolved components into a solid phase occurs only if the soil water is oversaturated with respect to the dissolved

components in the reaction. On the other hand if the soil water is undersaturated, and there exist a solid phase, dissolution of the solid phase will occur until

equilibrium is reached (Gustafsson et al., 2010). Formation of Al – and Fe- P are favored by low pH while Ca-P may form in very high pH (von Brömssen et al., 1985).

For sand filters in particular there are studies indicating that adsorption to Al (III)-(hydr)oxides together with precipitation into AlPO4 are one of the most important retention mechanisms for P (Eveborn et al., 2012b; Eveborn et al., 2014). Stuanes & Nilsson (1987b) also got results indicating Al was most important for P retention in their studied infiltration systems and sand filters. Regarding formation of Ca phosphates in sand filters Eveborn et al. (2012b) suggest this mechanism not to be of importance for long time retention of P in the majority of the sand filters in their study. Eveborn et al. (2014) found a significant amount of P bound in organic material in their studied filters. However this was not seen as an important mechanism for long term removal of P due to degradation of the organics.

(26)

organic matter. Decreased pH can enhance P sorption to Al/Fe(hydr)oxides and precipitation to Al/Fe phosphates and was seen in Eveborn et al. (2014) as an important mechanism of increased P removal capacity in sand filters. Also an increase of Al and Fe (oxalate-extractable Fe and Al) were evident in Eveborn et al. (2014) and Eveborn et al. (2012b) which was suggested to be due to the pH

decrease since it favors weathering and may therefore cause a release of oxalate-extractable Fe and Al to the soil (Eveborn et al., 2012a). Eveborn et al. (2014) and Eveborn et al. (2012b) also suggest this to be important for P removal in sand filters.

Precipitation or adsorption of phosphate are not irreversible reactions thus

adsorbed or precipitated phosphate can enter solution again. Eveborn et al. (2012b) suggest that P adsorbed to Al (III) /Fe (III)-(hydr)oxides as well as phosphates precipitated as Al-P or Fe-P can enter solution at times when P concentration in the influent water is low. P could for example be subject to wash out from the sand filter due to diluted wastewater or groundwater intrusion as shown in pilot scale column studies by Eveborn et al. (2014).

Technical description of the design of a conventional sand filter

To create an unsaturated flow regime in the sand filter wastewater can be applied to the infiltration surface in doses (a number of times per day) or in daily loadings (mm/day) (Siegrist et al., 2000). SEPA (1991) recommend a daily wastewater load to be between 50 and 60 L/m2_{/day (5-6 cm/day) if the sand is within their}

recommendations (figure 3). To facilitate creation of aerobic conditions in the sand filter the drainage pipes should be connected to aeration pipes (SEPA, 1991).

To facilitate infiltration it is recommended that a conventional sand filter is constructed with an application layer with a depth of 30-35 cm (figure 4). Application pipes should have a tilt of 5-10 promille to ensure distribution over whole surface if the system is gravity fed. For pressure systems no tilt is needed. The application layer should be composed of shingle or washed macadam with a grain size larger than the sand filter material but with largest 16-32 mm for gravity fed systems or 12-24 mm for pressure systems (SEPA, 1991) (SEPA, 2006).

Between application layer and the sand filter a transit layer of 3-5 cm with a finer fraction (4-8 mm) is recommended (SEPA, 1991) (SEPA, 2006). The depth of overlaying soil should be at least 40 cm (SEPA, 1991).

(27)

underlaying drainage layer a material separation layer of 5 cm with a grain fraction of 2-8 mm or 4-10 mm, depending on the size of the overlaying sand, is necessary (SEPA, 1991) (SEPA , 2006). Drainage pipes should have a tilt of 3 promille and the drainage layer should be composed of washed soil material (8-16 mm or 12-24 mm). It should be 20 cm deep if little or no percolation is desired and a larger depth, 30 cm, if it is desired (SEPA, 1991).

Figure 3: Cross section of a conventional sand filter constructed as one bed including depth of material layers as well as material type and size. Picture modified from SEPA (2006).

(28)

MATERIALS AND METHOD | 15

3 Materials and method

A literature study on treatment capacity of sand filters in previous studies was carried out as well as a field sampling campaign at two sites. In order to find out if the systems were working properly and if they met the recommendations from HaV concerning effluent quality and treatment performance a so called black-box (figure 5) study was carried out. Measurements and sampling of wastewater were only performed on the influent and effluent water and not of the bed material.

In the following section materials used in the study and preparations are presented. Furthermore the studied sites and the sand filters are described. Methods for influent and effluent sampling, on-site and lab measurements as well as lab analysis of the wastewater samples is also included.

3.1 Literature study on treatment capacity of sand filters

As mentioned in the scope the literature study was limited to the Nordic countries. Literature was searched for at the KTH library using the online search function KTHB Primo. Peer reviewed articles, an unpublished manuscript, institutional reports from e.g SEPA and HaV a degree project were included (table 2). Black box studies and mass balances studies for soil based on-site systems with a design capacity between 1 and 200 PE were included. A review of influent (sampled after septic tank), effluent and treatment capacity of tot-P, tot-N, NH4-N, DOC and BOD5 or BOD7 from the studies (average, minimum and maximum) were made. Most systems were conventional sand filters, but a sand filter with following P-filter as well as a filter with Filtralite was also included. Treatment capacity (TE) were in

Figure 5: The sand filter is seen as a black box. Water of a certain quality enters, is subject to processeses which are not studied and leaves the box with another quality. Soil and ground water intrusion as well as effective precipitation will dilute the effluent.

(29)

16 | MATERIALS AND METHOD

most cases taken directly from the studies but in some cases, where it was not reported in the studies, it was calculated from reported influent concentration (cin) and effluent concentration (cout) using equation 1. See table 2 for the included studies and a note on how treatment capacity presented in the literature study was obtained.

TE=cin−cout cin

⋅100 % (1)

Table 2: Studies included in the litterature study

Study Type of study Comment on which values are included in the literature study

Zhang & Renman (n.d)

Black box Nyberg & Pell (1983)

Black-box Black-box Black-box Black-box Black-box Liss (2003) Black box Black box Eveborn et al. (2012a)

Reported TE was used in the literature study

Mass balance Reported TE was used in the literature study

Reported average, minimum and maximum values of TE, inf and eff in the report was used in the literature study.

TE was calculated by the authours both with and without dilution. For the literature study TE and effluent without dilution which are presented in the report is used. However, reduction with dilution resulted in negative results. The authours did not include 2 of these 3 negative result when calculating average. In the literature study all results were included when calculating average.

von Brömssen et al.

(1985) Reported average, minimum and maximum values of TE, inf and eff in the _{report was used in the literature study.} Stuanes & Nilsson

(1987a) In the literature study TE:s calculated by the authours is used expect from _{BOD7 which I calculated from influent and effluent values presented in} report. Tot-N is calculated as kje-N+ NO3-N. In the literature study the results from four sand filters (Gröna Lund, Skurup, Hagen and Hovland) out of eight sand filters were used since the excluded ones were ponded, or were not real sand filters.

Aaltonen & Andersson

(1995) TE in report was calculated for summer and winter months. But for the _{literature study TE for all months was calculated from presented influent} and effluent concentrations. Average, minimum and maximum inf and eff were calculated from reported inf and eff.

Nilsson et al. (1998) In the report average, minimum and maximum TE as well as influent and effluent concentrations for all systems were presented without extreme values or negative TE. In this literature study these results were presented along with average, minimum and maximum values with extreme values and negative values included (which were calculated from data included in report).

Reported average, minimum and maximum values of TE, inf and eff in the report was used in the literature study.

Matamoros et al.

(2009) Reported average, minimum and maximum values of TE, inf and eff in the _{report was used in the literature study.} Black box pilot

scale column study Eveborn et al.

(30)

3.2 Meteorological data collection

Data on hourly air temperature and daily average precipitation from weather stations located within 25 km from each site were retrieved from SMHI (2015). Temperature data was needed to see if there had been a risk of frozen water in the sampler tubes during sampling and precipitation data was retrieved for a

discussion on possible dilution of the effluent water due to infiltrated rainwater. Data from ten days prior field sampling were retrieved for precipitation with respect to the retention time in the systems and temperature data were collected from the night time during the field sampling. Temperature data were collected from Stockholm-Bromma station for site 1 (7 km from site) and Landsort A for site 2 (ca 25 km from site). Precipitation data was retrieved from Norsborg II weather station for site 1 (7 km from site) and Wiad weather station for site 2 (15 km from site).

3.3 Study sites: Characterization of the systems

Both systems in this study are sand filters which follow primary septic tank

treatment (table 3). After filtration of the wastewater through the filter media it is directed to more extensive treatment, such as a P removal filter, or to a recipient. Since this study focus on the treatment capacity of the filter further treatment steps connected to the systems are not described.

Table 3: A comparison of the sand filters

Site 1: Ekerö Municipality Site 2: Södertälje Municipality

Year operation started 2012 Ca 1993

Households connected 13 37

Connected (PE)3 ₃₅ ₁₀₀

Effective infiltration area per bed (m²)

110 345

Hydraulic load (m3_/day)4 _{5.95 (5.25-7)} _{17 (15-20)}

Hydraulic surface load (cm/day)5 _{2.7 (2.3-3.2)} _{2.5 (2.2-2.9}

Surface load per application event6

(mm)

2.7 mm 1.5 mm

Recipient Mälaren Baltic Sea

Primary treatment Septic tank Septic tank

3. Based on the assumtion that the households are one family houses which own their house then average number of persons per househould is: 2.7 persons according to SCB (2016).

4. Based on template from HVMFS 2016:17 saying that wastewater volume per person and day on average is 170 L, with range 150-200 L. 5. Calculated as hydraulic load divided by total surface of beds which are in operation

(31)

More detailed descriptions of each system follow below. The information was retrieved mainly from construction maps but also from site visits and through contact with users and persons responsible for the systems. An assumption is made that the systems are built according to the construction maps. See Appendix for a comparison of both systems in a table.

Site 1: Sand filter in Ekerö municipality

Site 1 is located in Ekerö Municipality and has lake Mälaren as a recipient. The system, to which 13 households (35 PE) were connected when this study was carried out, was put into operation in 2012. The sand filter is located in postglacial clay (SGU, 2016) (figure 6) and the vegetation on the site consisted of grass.

An overview of the system can be seen in figure 7. It consists of two infiltration basins, each with a area of 110 m², which are alternately fed by primary treated wastewater from a pump well. The pump is triggered when the water reaches a threshold level in the pump well and aeration of the wastewater takes place prior to

Figure 6: Soil types at site 1 and approximate location of the sand filter (blue rectangle). Light yellow:

(32)

pumping the primary treated wastewater ca 40 m to the sand filter. Each sand filter basin have four application pipes and through them 2.7 L/m² (2.7 mm) of wastewater is distributed onto the filter surface per application event. The application is probably pressurized since no distribution well is installed before application pipes. Filtered water is drained from the system by a drainage pipe which is located perpendicular to the application pipes.

(33)

The sand filter is constructed with the surface of the application layer in line with the ground level and with application modules laying on the application surface (figure 8). Application layer bedding is composed of washed shingle (16-32 mm) and is 10 cm deep. Overlaying soil is located at minimum 30 cm above the application modules and application layer and overlaying soil is separated by a geotextile. There is no transit layer between the application layer and the filter. The filter media is a sand and has a depth of 60 cm. Below the filter there is a material separation layer with a size of 4-8 mm. The following drainage layer is 30 cm deep and is composed of shingle (16-32 mm). Drainage pipe is aerated (aeration pipe is connected to highest point of drainage pipe). Between the sand filter system and the surrounding soil there is a tight geotextile.

Site 2: Sand filter in Södertälje municipality

Site 2 is located in Södertälje municipality and the effluent from the system is released into the Baltic sea. The system was put into operation around 1993 and 37 households (100 PE) were connected at the time of this study. It is located in a postglacial fine clay (SGU, 2016) (figure 9) and is overgrown by high vegetation.

(34)

Pretreatment consist of two septic tanks with a total volume of approximately 29 m³. With a daily hydraulic load of 15-20 m³/day (table X) the theoretical hydraulic retention time in the septic tanks is 1.4 - 1.9 days.

An overview of the system design can be seen in figure 10. The system consist of three infiltration basins, with an area each of 345 m². Two are alternately operated and one serves as a substitute. Wastewater is pumped ca 700 m from the pump well (the pump is also in this case triggered when water reaches a threshold level) to a distribution well from which the wastewater is distributed to the application pipes by gravitation.

Wastewater is distributed in seven application pipes (10 ‰) per infiltration basin, with a surface load of 1.5 L/m² (1.5 mm/day) per application event and filtered through the filter. Water is drained from the system in eight drainage pipes (with a tilt of 10 ‰) per infiltration basin which are located parallel to the distribution pipes. The drainage pipe join together in one drainage pipe at the end of the bed (5 ‰) which transports the water towards another pipe which goes to the recipient.

(35)

The sand filter system is covered by a layer of clay which is 80 cm thick at

minimum (figure 11). There is a material separation layer between the application layer and the overlaying material which is composed of geotextile. The application layer is 25 cm deep and composed of washed shingle (16-32 mm) and there is no transit layer below the application layer. The filter layer is 80 cm deep and consists of a sand with a grain size distribution within the recommendation from the SEPA (2006).

(36)

There is no material separation layer between the sand filter and the drainage layer, which is composed of washed shingle (16-32 mm) and is 20 cm thick. The highest point of the drainage pipes are connected to an aeration pipe. If the bottom is separated from surrounding soil with water tight layer is not known from construction map.

Organic micropollutant removal in the sand filters at site 1 and site 2

Influent and effluent wastewater samples obtained in this study have also, as mentioned before, been subject to analysis for organic micropollutatns within the FORMAS project RedMic (i.e Blum et al., 2017). They made an analysis of 26 target micropollutants (MP:s) belonging to the following groups: biocides, food additives, fragrances, linear alkyl benzenes, organophosphorus flame retardants, plasticiser, polymer impurity, rubber additives, surfactants and UV stabilisers. The median treatment efficiency, with respect to all included MP:s was 85 % for site 1 and 83 % for site 2 (table 4) which were the highest among the studied on-site systems (5 in total).

Blum et al. (2017) found a significant correlation between hydrophobicity for the MP:s, measured as log(KOW), and median removal efficiency in all systems incuded in their study (both on-site and treatment plants). As explanied in their study hydrophobic MP:s have higher affinity towards organic material and therefore

(37)

more likely to remain in the sand filter compared to less hydrophobic MP:s. Some MP:s (e.g TMDD and TDCPP) were, however, overall removed worse than expected based on their hydrophobicity.

Table 4: Selected data from Blum et al. (2017) on organic MP:s removal in the sand filters at site 1 and site 2

Micropollutant Group TE site 1 (%) TE site 2 (%)

Triclosan (TCS) Biocides 98 91

α-tocopheryl acetate (α-TPA)

Food additives 96 94

Galaxolide (HHCB) Fragrances 95 96

Octocrylene (OC) UV stabiliser 99 98

Tributylphosphate (TBP) Organophosporus flame retardants 72 83 Benzophenone (BP) UV stabilisers 85 72 2-(Methylthio) benzothiazole (MTBT) Rubber additives 72 64 2,4,7,9-Tetramethyl-5-decyn-4,7-diol (TMDD) Surfactants 47 31 Tris(1,3-dichloro-2-propyl)phosphate (TDCPP) Organophosporus flame retardants 56 83 Median (26 MP:s) 85 83

3.4 Field measurements and water sampling

This section describes preparations and materials used, approaches used for water sampling in the field as well as sample handling and storage.

Materials and sampling preparations

Two programmable automatic water samplers with peristaltic pump, namely Teledyne Isco 6712 and Teledyne Isco 2900, were used for the field sampling. Old silicone pump tubing was used for both samplers and a new PVC suction line. Pump tubing and suction lines were cleaned with isopropanol (25%) and distilled water prior field sampling at each site to avoid crosscontamination.

(38)

volume with an accuracy which differs with ±10% (Isco 6712) and ±10ml (Isco 2900) of programmed nominal value.

The automatic samplers were programmed to take a 24 h composite sample composed of time proportional samples taken every hour. An assumption that the septic tank levels off large variations in incoming flow due to long HRT according to Palm et al. (2002) was made in order to justify to take time proportional samples instead of a flow proportional samples. On the other hand since the sand filter is fed with wastewater in doses outgoing water is likely to vary. However, an

assumption was made that the variations were small due to limited possibilities of getting hands on a flow meter. The automatic samplers were calibrated in the field to take an hourly sample with a nominal volume around 130 ml (see Appendix). Grab samples were taken at influent and effluent at each site as well for comparison to composite samples. For measurement of DO, wastewater temperature and pH in the field Hq40d multi from HACH was used.

Wastewater sampling at site 1

Water sampling at site 1 was carried out in early November 2015 (Wednesday-Thusday). A grab sample and a 24 hour composite sample was taken for effluent and influent respectively (see Appendix for details). DO levels, pH and

temperature was measured in both influent and effluent on site at the same time as the grab water sample was taken.

Wastewater sampling at site 2

Wastewater sampling at site 2 was carried out in the middle of November 2015 (Monday-Tuesday). DO levels, pH and temperature was measured on-site in both effluent and influent when sampling was initiated at the same time as the grab sample was taken (see Appendix for details).

Daily wastewater load

Daily wastewater load to the system at site 1 was not practically possible to estimate with the available resources for this study. However at site 2 the pump which pumped wastewater from the pump well to the sand filter system had a clock which registered the number of hours the pump had been operating. Therefore the time the pump was active during the sampling (t24h) could be achieved. On-site a manual

inflow measurement (qin) was made and the time it took for the pump to empty the

(39)

height between the upper and lower threshold level in the well was measured in order to calculate the volume of water between the threshold levels (V). Total outgoing volume during the pump event (Vout) was thereafter calculated (equation

3). The volume the pump is pumping per time unit (q) was calculated (equation 4) and thereafter total volume which was pumped to the sand filter system during the sampling period (V24h) was determined (equation 5) . See Appendix for calculations.

V_in=q_in⋅t (2)

V_out=V_in+V (3)

q=Vout

t (4)

V_24h=q⋅t_24h (5) Sample transport and handling

After the sampling period the glass bottles with the composite samples were stored in freeze boxes, transported to lab and divided into smaller plastic bottles which were refrigerated. Prior to lab analysis the samples moved to room temperature to melt.

3.5 Lab analysis

This section briefly describes the method for lab analysis of the wastewater

samples. Physio-chemical parameters including turbidity, DOC, tot-P and NH4-N were analysed.

Turbidity

Turbidity was measured in the lab using ISO Turbidimeter 2100P from HACH. Dissolved organic carbon

Prior analysis the samples were filtered through a membrane filter with a 0.45 µm pore size in order to remove solid fractions. Subsequently the samples were

(40)

heating the sample to 680 °C together with oxygen as an oxidant (and a platinum catalyst) whereby all carbon is combusted to CO2 which is detected by the NDIR. Total organic carbon, or in this case DOC since the sample was filtered, is

subsequently calculated as DOC = TC – IC. Ammonium N

Ammonium N (NH4-N) was determined using AA3 SEAL AutoAnalyser according to ISO 11732. In short the sample reacts with salicylate and dichloroisocyanuric acid to form a blue compound. The absorbance of the sample is thereafter measured at 660 nm with a colorimeter. (SEAL Analytical, 2011)

Total phosphorus

(41)

RESULTS | 29

4 Results

In this section results from the litterature study on treatment capacity of sand filters are presented followed by meterological observations during and prior to samplig, including precipitation and air temperature. Estimated daily wastewater load to the system at site 2 is also presented in this section. Influent and effluent concentrations of DOC, NH4-N and tot-P from both sites are presented together with on-site measurements of pH and temperature as well as lab measurements of turbidity in influent and effluent samples.

4.1 Literature study on treatment capacity of sand filters

There exist a limited number of studies which evaluates the function of sand filters in the nordic countries (e.g Nyberg and Pell, 1983; Stuanes and Nilsson, 1987; Aaltonen and Andersson, 1995; Nilsson, Nyberg and Karlsson, 1998; Liss, 2003; Malmén and Jönsson, 2002; Matamoros et al., 2009; Eveborn et al., 2012b; Zhang & Renman, n.d). The majority are from the 80ties and 90ties and most of them are black-box studies in which treatment efficiency (TE) is calculated from sampled influent (cin) and effluent (cout) concentration (equation 1). Among the black box studies grab samples were taken in some (e.g Zhang & Renman, n.d; Matamoros et al., 2009; Nilsson et al., 1998) and in others time proportional (e.g Liss, 2003) or flow proportional (e.g Nyberg & Pell, 1983) composite sampling was carried out. Mass balance studies on sand filters, in which P content of the bed material is analyzed and compared to an estimation of the total P load during its lifetime, are less common (e.g Eveborn et al., 2012b; Eveborn et al., 2014 ). See Appendix for comparison of average, minimum and maximum influent concentration, effluent concentration and TE from the studies included in the literature study.

(42)

30 | RESULTS

accounting for dilution for tot-P). Nilsson et al. (1998) concludes that dilution was about 60 % for large sand filters in their study, but this is not accounted for when calculationg TE.

Nilsson et al. (1998) get negative reduction of tot-P as well as tot-N but decides to not include theses values when calculating average TE for their sites. Average TE for tot-P is 65 % when excluding negative values and 46 % when including them. For tot-N the difference is not as large (59 % when excludning and 52 % when including).

Capacity of N and P removal in previous black-box studies

Treatment efficiency for tot-P and tot-N in previous black box studies varies a lot. Some studies report treatment efficiencies for tot-P ranging from negative

reduction up to almost 100 % (e.g Zhang & Renman, n.d; Aaltonen & Andersson, 1995; Nilsson et al., 1998) indicating that the TE is highly variable among sand filters. Avergage TE also varies in the litterature with most having an average reduction between 45-80 % (e.g Zhang & Renman, n.d; Nyberg & Pell, 1983; Stuanes & Nilsson, 1987a; Aaltonen & Andersson, 1995; Nilsson et al., 1998; Liss, 2003).Treatment efficiency for tot-N is also variable. Some studies report

negative values (e.g Aaltonen & Andersson 1995, Nilsson et al., 1998) up to 70 or 80 % (e.g Zhang & Renman, n.d; Stuanes & Nilsson, 1987a; Nilsson et al., 1998). Most studies have an average TE for tot-N between 28-60 %. Effluent

concentrations for tot-P in the studies have an average between 3-10 mg/L, most being around 3 or 4 mg/L. Average tot-N effluent concentration in the studies was between 20 and 70 mg/L, most being 30-40 mg/L.

In well functioning systems usually almost all NH4+ is nitrified (Siegrist et al., 2000) but since not all systems are properly functioning Palm et al. (2002)

conclude that around 0-90 % of NH4+ is removed in a sand filters in the literature. As an example of the variations between sand filters Matamoros et al. (2009) reported an average reduction of 76 % in two sand filters, with reduction ranging from 46-99 %. In Liss (2003) 72 % was nitrified on average and in Stuanes & Nilsson (1987a) 34 % was nitrified on average.

(43)

RESULTS | 31

dimensioned with a surface area larger or equal to 5 m2_{/pe have a capability of} reducing BOD 85-97 % if properly functioning and designed. I a column study in Eveborn et al. (2012a) with material from four sand filters were BOD reduced by 100 %. There exist a few studies which includes DOC as a parameter. In Zhang & Renman (n.d) DOC was reduced on average 50 %, but the TE ranged from negative up to 74 %. Effluent DOC concentrations were on average 17 mg/L.

Treatment capacity of tot-P from mass balance studies

Eveborn et al. (2012b) carried out a mass balance study on a open filter bed. P accumulation was estimated from samples of the sand material and compared to total P load. The study concluded that 12 % ± 4 % of total influent tot-P amount added to the sand filter during its life time had accumulated in the bed. It should be noted however that the open sand filter in the study was designed for 225 PE and had a design hydraulic loading rate of 33 cm/day which is high compared to what is recommended by the SEPA for conventional sand filters.

Eveborn et al. (2014) sampled bed material from six filter beds of which two were open (one of them was the same as in Eveborn et al. (2012b). The sand filters were of different size designed for 5-225 PE. The study concludes that treatment

efficiency of tot-P in the long run depends a lot on the magnitude of the hydraulic loading rate to the system. The sand filters could theoretically, given the tot-P concentration in the influent to be 10 mg/L and a soil treatment volume of 1 m3_per m drainage tube, accumulate P during 3-8 years if surface load is 3 cm/day but thereafter they would be saturated with respect to P. With a lower hydraulic

loading rate they could last longer until saturation and the other way around with a higher hydraulic loading rate.

4.2 Metrological observations and daily wastewater load

(44)

32 | RESULTS

4.3 Quality of influent and effluent

The temperature of the influent was 12 ºC at both sites. Effluent temperature was 11 ºC for site 1 and 9 ºC for site 2. Dissolved oxygen levels in pump well (influent) was 1 mg/L at site 2 whereas 2.6 mg/L at site 1. After aeration DO levels increased up to 7.4 mg/L at site 1. In effluent water DO levels were 8.5 mg/L at site 2 and 4.2 mg/L at site 1 (table 5). pH was lower in effluent water compared to influent water for both sites (table 5).

Table 5: pH and dissolved oxygen level in influent and effluent at site 1 and site 2

pH Dissolved oxygen (mg/L) Site 1 Site 2 Site 1 Site 2

Influent before aeration 7.1 6.9 2.6 1.0

Influent after aeration - - 7.4

-Effluent 6.3 6.4 4.2 8.5

Samples from site 1 had on average 3 times higher turbidity in influent compared to site 2 whereas site 2 effluent had a higher turbidity on average (figure 14). Grab samples had higher influent values compared to composite for both sites. Composite samples had a turbidity in the influent of 60 NTU for site 2 and 200 NTU for site 1. Turbidity in the composite effluent samples were 16 NTU and 5 NTU for site 2 and site 1 respectively.

0 2 4 6 8 10 12 14 Date P re ci pi ta tio n (m m )

(45)

RESULTS | 33

Influent DOC concentrations were on average almost 2 times higher for site 1 compared to site 2 (figure 15). Composite and grab samples for site 1 have similar influent DOC concentrations as well as effluent DOC concentrations. For site 2 influent concentrations for grab (50 mg/L) and composite sample (47 mg/L) are similar. Concentrations in composite samples from site 1 were 98 mg/L and 14 mg/L for influent and effluent respectively.

NH4-N influent concentrations was on average 3.6 times higher for samples from site 1 compared to samples from site 2 (figure 16) . Whereas effluent concentrations

Figure 13: Turbidity (NTU) in composite sample (C), grab sample (G) at site 1 and site 2

Site 1: C Site 1: G Site 2: C Site 2: G 0 50 100 150 200 250 Influent Effluent

Figure 14: DOC concentration (mg/L) in composite sample (C) and grab sample (G) at site 1 and site 2. *no data

n.d*

(46)

34 | RESULTS

were 2.7 higher for site 1 samples compared to site 2 samples. Composite and grab samples had similar influent and effluent concentrations for both sites. Composite influent concentrations were 24 mg/L and 88 mg/L for site 2 and site 1 respectively while composite effluent concentration was 2.0 mg/L for site 2 and 4.9 mg/L for site 1.

Tot-P concentration in influent was on average 2.8 times higher for site 1 compared to site 2 and concentration of tot-P in effluent was on average 3.3 times higher in site 1 samples (figure 17). The concentration of tot-P in influent and effluent was similar for composite contra grab samples for both sites. Tot-P influent

concentrations in composite samples were 2.9 mg/L and 8.0 mg/L for site 2 and site 1 respectively while effluent concentration was 1.3 for site 2 and 4.6 for site 1.

Figure 15: NH4+concentration (mg/L) in composite sample (C) and grab

sample (G) at site 1 and site 2

Site 1: C Site 1: G Site 2: C Site 2: G 0 10 20 30 40 50 60 70 80 90 100 Influent Effluent

Figure 16: tot-P concentration (mg/L) in composite sample (C) and grab sample (G) at site 1 and site 2

(47)

DISCUSSION | 37

5 Discussion

In this section mechanical, biological and chemical function of the sand filters are discussed based on parameter values or concentrations in influent and effluent. Possible dilution of the wastewater before entering the sand filter is discussed based on comparison with expected incoming concentration of P and the actual wastewater load (only for site 2). Further down the sampling procedure is

discussed including possible shortcomings, uncertainties and the reliability of the results.

5.1 Diluted influent to the sand filters

Influent concentrations as well as estimated wastewater load to site 2 suggest that influent to the system at site 2 is diluted. Influent concentrations, in grab as well as composite samples, of tot-P, NH4-N, DOC for site 1 are at least 2 times influent concentrations for site 2. Also turbidity influent measurements for site 1 are significantly higher compared to site 2. Expected tot-P influent concentrations from HaV (HVMFS 2016:17), based on templates on average daily water usage per person (170 L/person/day with a range of 150-200 L/person/day) and daily load tot-P per person, is 12 mg/L with a range of 5-15 mg/L depending on water usage. Influent concentration for site 1 (7.96 mg/L) is within the expectations whereas influent from site 2 (2.87mg/L) is below the expected influent concentration which suggests that influent water is diluted.

Furthermore, estimated wastewater load to the sand filter during sampling at site 2 represents a water load of 330 L/person/day (given 100 PE connected). This is above the range of assumed personal water consumption in HaV (HVMFS 2016:17) of 150-200 L/person/day on average 170 L/person/day. It also indicates that water is diluted before entering sand filter. Meteorological observations from the time before sampling show recorded rain events within 10 days prior sampling which might have resulted in soil water or groundwater seeping in to wastewater pipes. It should be noted that connected PE is an estimation based on number of

(48)

38 | DISCUSSION

more PE are actually connected, say 165 PE, then daily wastewater load per person would be 200 L/person/day and thereby within the range of expected wastewater usage from HaV. However, since the influent tot-P concentration is much lower than tot-P expected concentration from HaV results suggest that influent at site 2 is diluted.

5.2 Biological and mechanical function of the sand filters

Effluent ammonium (NH4+) concentration is, compared to influent, concentration very low for both sites. Calculating treatment efficiency yields 91 % reduction both grab and composite sample for site 2. For site 1 the reduction was slightly higher with 95 % for composite and 92 % for grab sample. This indicates good biological function when it comes to nitrification in both systems and hence that the systems are well aerated. Also a pH decrease is evident at both sites which could be due to nitrification as hydrogen ions are released during the process of nitrification (Wiesmann et al., 2007). Also pH at both sites are within the pH range for the nitrifying bacteria to function (5.8-8.5) (Wiesmann et al., 2007). Denitrification in the systems, as well as reduction of tot-N, is unknown since nitrites and nitrates was not analysed for.

Measured level of dissolved oxygen in effluent water also indicates that the aeration is well functioning in the systems as it is 4.2 mg/L for site 1 and 8.5mg/L for site 2. Similar oyxgen levels to site 1 in effluent from sand filters are reported in

Matamoros et al. (2009) whereas site 2 had a higher oxygen level in comparison. Effluent water at site 2 is almost 80 % saturated with respect to oxygen as oxygen saturation at 12 °C is 10.8 mg/L (Vesilind, Peirce & Weiner, 2013).

As aerobic conditions favor biodegradation of organic substances the results suggest that the biological function when it comes to reducing BOD is good as well although this parameter was not analysed. Removal of DOC could give a slight indication on biodegradation of organic substances in the systems although there is also suspended organics present in the wastewater which are included when BOD is measured. Removal of DOC was 68 % for site 2 grab, 86 % for site 1 composite and 87 % for site 1 grab which all are above the average removal rate in Zhang &

(49)

DISCUSSION | 39

Turbidity in composite samples is reduced by 73 % at site 2 and 97.5 % at site 1 which indicates that organic and inorganic suspended solids are reduced as well since suspended solids contributes to turbidity. The exact reduction of suspended solids is however not possible to say from turbidity measurements since correlation between turbidity and suspended solids concentration is not established for the wastewater at the sites.

5.3 Chemical function of the sand filters

Removal rate for tot-P in the systems are 42 % for site 1 and 54 % for site 2. Which both are similar to average in the majority of the studies in the literature (40-80%) . Site 1 has a effluent concentration (4.6 mg/L) within the range of averages in the literature (3-10 mg/L) whereas site 2 has an effluent concentration (1.3 mg/L) which outside the range.

None of the systems lives up to the reduction level for normal or high protection level from HaV when it comes to tot-P. If effluent concentration is considered instead of reduction then site 2 fulfills the requirement from HaV for normal protection level. However, since the influent water was most likely diluted, site 2 is not considered to meet the requirements for normal protection level.

5.4 Effects of variations in influent concentration on reliability of results The water which was sampled at the inlet is not the same water which was sampled at the outlet if the delay between the start of influent and effluent sampling isn’t equal to the HRT of the system. If influent concentration show hourly and/or diurnal variations then calculating TE from the influent and effluent concentration does not yield an accurate value. HRT for the studied systems are unknown and were not taken into account in this study but as an example has HRT in sand filters has been reported by Nilsson et al. (1998) to be 6 days on average in the sand filters in their study, ranging between 5 and 11 days. Also Ejhed et al. (2012) did a tracer test with bromid and found that HRT in a system with biomodules, sand filter and a P-filter was 2 days. It can be compared to the delay between the start of sampling at inlet and outlet in this study which was 2.5 hours and 1 hour for site 1 and site 2 respectively. Thus the effluent water which was sampled might have entered the system say 2 or 6 days before start of sampling and could have a different

parameter composition compared to the sampled influent. Therefore there is a uncertainty when it comes to the estimated TE of the sites. If influent concentration show low variations the result could be used as an approximate value of the

(50)

40 | DISCUSSION

To account for possible daily, weekly and even seasonal variations in influent concentrations, and thereby get a more accurate picture of the reduction in the systems, sampling could be conducted during a larger period of time. Also if sampling is undertaken during a longer period of time the effect of HRT will not impact the result as much.

5.5 Effect of variations in flow on reliability of results

In this study time proportional sampling was performed due to limited resources and it can be discussed whether flow proportional sampling would have given a composite sample with a better representation of reality. Especially since outgoing water would probably show variations in flow due to the fact that the systems were fed in doses resulting in an uneven outgoing flow.

5.6 Effect of dilution of effluent from precipitation, soil water and groundwater

Another uncertainty in this study is the magnitude of dilution of effluent from groundwater intrusion or effective precipitation. Dilution would cause lower effluent concentrations and thereby a possible overestimation of the TE if

magnitude of the dilution is significant compared to the wastewater load. Nyberg & Pell (1983) estimated the dilution by a tracer test and concluded that the effluent was diluted 2 to 4 times. Nilsson et al. (1998) also measured chloride concentration and concluded that effluent water was dilluted around 60 % for the large systems in their study. Ejhed et al. (2012) made a tracer test with bromide and effluent

concentations were lower compared to influent concentration which they thought was due to dilution, dispersion and/or adsorption of bromide to organic matter in the bed.

On a yearly basis dilution from effective precipitation, calculated as actual precipitation (SMHI, 2014a) minus evapotranspiration (SMHI, 2014b), is on average 20 cm/year at maximum. This is if all effective precipitation would percolate through the clay and into the sand filters. Compared to the average hydraulic load per year (986 cm/year for site 1 and 913 cm/year for site 2) it is small and consequently dilution on a yearly basis is 2 % for site 1 and 2.2 % for site 2. However, since the sand filters are surrounded of clay, which has a very low hydraulic conductivity, only a small share of the effective precipitation is likely to infiltrate resulting in a lower average yearly dilution.

(51)

DISCUSSION | 41

were no precipitation during or 10 days prior the sampling period. At site 2 however, sampling took place during a wet period and therefore there is a

possibility that the effluent was diluted by effective precipitation. However, since the sand filter is located below a 80 cm thick clay layer surrounded by clay the share of effective precipitation which infiltrates into the sand filter probably is low due to clay having low hydraulic conductivity, the same for the level of groundwater and soil water intruision. At site 1 the sand filter also is located in clay, below a layer which is at least 30 cm above top of application modules which also indicates low groundwater and soil water intruision and infiltration by effective precipitation during wet periods.

5.7 Uncertainties about long term treatment efficiency of P

Black box studies have been criticized in recent years when it comes to determining TE of tot- P. Eveborn et al. (2012b) suggest that black box studies have been, and are, overestimating treatment efficiency for sand filters when it comes to tot-P and that the sand used in sand filters cannot retain P in the long run as the

P-adsorption capacity of the material not is high enough (Eveborn et al., 2012b). As studies suggest (e.g Eveborn et al., 2012b; and Eveborn et al., 2014), phosphate is not irreversibly bound to in the sand material and can be washed out from the sand filter at times when phosphate concentration in incoming wastewater is low. Thus tot-P in effluent is likely to vary in time depending on phosphate concentration in influent. It is therefore reason to be skeptical to the results on reduction of tot-P for the sites 1 and site 2 in this study. In case of low variations in influent

concentrations as well as minor dilution of infiltrated wastewater it can represent a snap-shot of tot-P reduction but probably not the long term reduction in the

Function of soil-based on-site wastewater treatment systems - Biological and chemical treatment capacity

Function of soil-based on-site

wastewater treatment systems -

Biological and chemical treatment

capacity

LINNÉA TJERNSTRÖM

Function of soil-based on-site

wastewater treatment systems

- Biological and chemical treatment capacity

LINNÉA TJERNSTRÖM

S

ummary in Swedish

Acknowledgements

Table of Contents

List of abbreviations

List of explanations of expressions and words

1 Introduction

2 Theoretical background

3 Materials and method

4 Results

5 Discussion

_{ummary in Swedish}