c:>soo ISSN 0169-6246 On the Design of Percolation Facilities Communication of the Department of Sanitary Engineering and VI/C'ltflr Management May 1996 fr. N.J. Monster I ir. M.J. Leeflang
Ni Faculty of Civil Engineering and Geosciences Department of Water Management, Environmental and Sanitary Engineering TU Delft Land- and Water Management Section Delft University of Technology On the Design of Percolation Facilities
Communication nr.: 66
M.J. Leetlang N.J. Monster
Delft University of Technology May 1996 Faculty of Civil Engineering Department of Water Management, Environmental and Sanitary Engineering Contents
Contents
Contents
Summary
Samenvatting V
1. Introduction
2. Applicable disconnection methods and their design 2.1 Introduction 3 2.2 EvaluatiOn of the various disconnection methods 3 2.3 Evaluatton of the design procedures 3 2.4 Shortcomings and proposed improvements for the design procedure 5 2.5 Proposed disconnection methods 5
3. A design graph based on simulations 3.1 Introduction 7 3.2 The traditional Storage-Design Discharge-Frequency curve 7 3.3 Development of the new des1gn graph 10
4. The modelling of the percolation trench, calculations and results 4. 1 Introduction 13 4.2 The reservoir model 13 4.3 Four different scenarios for the percolation rate from the reservoir 15 4.4 Simulations with the four scenarios 1 7 4.5 S1mulations with other scenarios 19 4.6 A percolatiOn trench with a drain on the bottom 20
5. Design graphs percolation trench 23
6. The Mulden-Rigole system 6.1 Introduction 33 6.2 The model 33 6.3 The optimal use of the system 34 6.4 The design graphs of the Mulden-Rigole system 35
7. Discussion, Conclusions and Recommendations 7.1 Discussion 37 7.2 Conclusions 38 7. 3 Recommendations 38
References 41 The Destqn of lnflltrauon and Percolauon F;~cthltes
Appendix A. Simulation of the precipitation-runoff process A.1 Introduction 43 A.2 RestrictiOns and Requtrements 43 A.3 The loss process 44 A 4 The delay process 46 A.5 The combtned loss and delay model 49 A.6 Output 50 A. 7 Runoff coefficient 52
Appendtx B. Extreme value analysis B. 1 Introduction 55 B.2 Theoretical background 55 B.3 The probabtlity dtstributton of partial duratton senes 55 B.4 The use of the parttal duration serres tn our model 56
Appendix C. Tundern 59
Appendtx D. The design of the Ruwenbos MR-system 65 Summary
Summary
The disconnection of impervious areas is a measure to reduce the overflow fre quency and hydraulic load of combined and improved separate sewer systems. Disconnection can be realized by guiding the storm water runoff to an mfiltrat1on and/or percolation facility. The des1gn of such facilities has been examined in this study.
Design standards and criteria from various countries have been studied and com pared. This showed that the current des1gn procedures ( = the determination of the requ1red storage volume of a facility) exh1b1t a number of spurious assump tions. Examples of these are: an mcorrect use of Intensity-Duration-Frequency curves; neglect of the loss and delay processes which occur during storm water runoff; assumption that discharge from a facility is constant in time.
The objective of this study is to remove the mentioned weak points from the current design procedures in order to design disconnection facilities in a more realistic way. A computer program has been written which simulates the infiltration-percolation process of a facility. The results of several simulations have been used for the construction of a design graph.
Storage-Design Discharge-Frequency curves are often used in the present design procedures. If the discharge from the facility is considered as a dynamical variable, the construction of this graph is no longer possible. A new design graph has been developed in order to solve this problem. In this new graph the permeability of the soil (K-value) is plotted against the required construction length (L-value). scaled to the amount of connected impervious area. For each graph the width and depth of a disconnection facility are chosen beforehand. For a known K-value and amount of impervious area the required construction length can be assessed directly with this graph.
A reservoir model was used to simulate the rainfall-percolation process. The input of the model exists of a data series of five-minutes-depths of precipitation over the period 1970 - 1984 as recorded in Lelystad. A separate loss and delay model IS applied to transform series of ra1n f1gures mto inflow figures. The calculated storage amounts are submitted to an extreme value analysis (part1al duration).
To determine the most suitable simulation model for the design graph, various alternative simulation models have been tested. The results have been compared in order to analyse the influence of each alternative on the required dimensions of the facility. F1rstly, it turned out to be save to construct the design graphs using a clogged bottom scenario. Secondly, it is recommended to use inflow data as input, and to cons1der the percolation d1scharge from the facility as a dynamical variable. As the effect of 'consecutive storms' results in an increase of the re-
iii The Oes•gn of lnflltrauon and Percotauon Fac•llt•es
qu~red d1mens1ons, 1t IS recommended to mclude th•s effect m the s•mulations as well. This is done by the use of a continuous mflow series rather than applymg des1gn storms.
Next. the effect of a discharge drain on the bottom of a percolatiOn trench is invest•gated. lt appeared that, compared to a fac•lity w•thout a dram, the reqUired dimens•ons of a facility with a drain decrease s•gnif•cantly. If the capac1ty of the drain increases, the permeability shows less influences on the requJred dimen Sions. Consequently so1ls w•th low permeability allow the applicatiOn of percola tion fac•hues, prov1ded a discharge drain is installed at the bottom of the facility.
The Mulden-Rigole system is a combination of a percolation trench w1th a drain and a surface storage. Simulations with this facility have been performed in order to examine the optimum use of the storage above and beneath the surface. lt turned out that the storage above surface is decisive for the design. The perme ability of the surroundmg soil appeared to have no significant influence on the required dimensions.
The construction of relevant design graphs (KL-graphs) is based on the results of the modelling as mentioned above. Various graphs have been constructed for the design of the following facilities each with different cross sect1onal dimensions: percolatiOn trenches w1th and without a dram. and the Mulden-Rigole system. These graphs are applicable to Dutch climate and soil cond•tions. provided the followmg aspects are considered. Firstly. the geo-hydrological aspects were s•mulated in a s1mphfied way. Secondly, the graphs are developed from a 'dis charge' point of v1ew.
iv Samenvattmg
Samenvatting
Het afkoppelen van verharde oppervlakken is een maatregel om de overstortings frequentie van een gemengd en een verbeterd gescheiden rioolstelsel te reduceren en de hydraulische belasting van het rioolstelsel te beperken. Door het afstromen de regenwater te leiden naar een infiltratie- en/of percolatievoorziening - al dan niet voorzien van een drain onderin - kan de afkoppeling warden gerealiseerd. Het ontwerp en de dimensionering van dit soort voorzieningen is m dat onderzoek nader bekeken.
In een literatuurstudie zijn ontwerpnormen en -regels uit binnen- en buitenland bestudeerd en vergeleken. Uit deze studie bleek dat de dimensioneringsprocedure ( = het bepalen van de benodigde berging van een voorziening) een aantal zwakke punten vertoont. In dit verband warden genoemd: Een onjuist gebruik van regen duurlijnen; het verwaarlozen van optredende verliezen en vertragingen bij het afstromingsproces; het constant veronderstellen van de afvoer uit de voorziening.
Doel van dit onderzoek is om genoemde zwakke punten te ondervangen en zo tot een meer realistische dimensionenng van afkoppelingsvoorzieningen te komen. Daartoe is een computer programma geschreven waaran het neerslag-infiltratie percolatie proces van een voorzaening wordt gesamuleerd. Het programma is gebaseerd op een bakmodel. Als invoer wordt een 15-jaar lange reeks vtjf-minuten sommen van de neerslag gebruikt. Een apart verlies- en vertragingsmodel vormt de regencijfers om tot inloopgegevens. De berekende geborgen hoeveelheden warden onderworpen aan een extreme waarden analyse. Met de resultaten van verschillen de simulatie's wordt vervolgens een ontwerpgrafiek geconstrueerd.
In de huidige procedures wordt veelal gebruik gemaakt van bergings-afvoer-grafie ken. Wordt de afvoer echter als dynamische variabele beschouwd. is de construe tie van deze grafiek niet meer mogelijk. Een nieuwe ontwerpgrafiek is ontwikkeld om dit probleem te verhelpen. In deze grafiek is de doorlatendheid van de bodem (K-waarde) uitgezet tegen de benodigde constructielengte (L-waarde) per hoeveel heid aangesloten verhard oppervlak. De breedte en diepte van een afkoppelings voorziening zijn voor iedere grafiek van te voren gekozen. Zijn de doorlatendheid en de hoeveelheid aan te sluiten verhard oppervlak bekend, kan de benodigde lengte van de voorziening direct met de grafiek warden bepaald.
Om het beste rekenmodel voor de dimensionering te bepalen, zijn met een aantal verschillende modellen simulaties gemaakt en zijn de resultaten met elkaar vergele ken. Er is geanalyseerd wat de invloed is van het betreffende rekenmodel op de benodigde afmetingen van de voorziening. Om aan de veilige kant te blijven war den ontwerpgrafieken geconstrueerd voor een percolatiesleuf waarvan de bodem is dichtgeslibd. Bovendien wordt aanbevolen om met inloop gegevens te rekenen en de afvoer als een dynamische variabele te beschouwen. Als wordt gerekend
V The Desagn ot lnhltrauon and Percolation Facohtael>
met een continue reeks kan het voorkomen dat de bergmg van een voorzienmg nog niet geheel leeg as op het moment dat de volgende bui begint. Dit verschaJnsel wordt de opeenvolgmg van bua en genoemd. Geadvaseerd wordt d•t effect ook mee te nemen aangezien het le1dt tot grotere benod1gde afmet•ngen.
Vervolgens is het effect van een afvoerdram op de bodem van een percolataesleuf onderzocht. De belangrijkste conclus1e is dat het toepassen van de drain leidt tot significant kleinere benodigde afmetmgen in vergehjking met een voorzaening zonder drain. Tevens bleek dat b11 een toenemende capac1te1t van de drain, de doorlatendheid van de bodem steeds mmder mvloed heeft op de benodigde afme tingen. Grondsoorten met een kleme doorlatendheid hoeven dus geen belemmering te vormen voor het toepassen van een percolat1esleuf, mits een afvoerdram wordt aangelegd op de bodem van de sleuf
Het Mulden-Rigole systeem is een voorz1enmg waarin een infiltrat1esleuf met drain 1s gecombineerd met een bovengrondse bergmg. Aan de hand van simulat1es met deze voorz1emng is de opt1male benutting van boven- en ondergrondse berging bekeken. Utt de stmulatte's volgde dat de bovengrondse bergmg maatgevend is b1j het ontwerp van deze voorz1ening. De doorlatendhetd van de bodem bleek geen SIQOif1cante invloed meer te hebben op de benod1gde afmettngen van de voome nmg.
Op basis van bovenstaande bev1ndmgen Ztjn verantwoorde ontwerpgrafieken (KL grafieken). met elk verschillende dwarsdoorsneden, voor percolatiesleuven met en zonder drain en het Mulden-R1gole systeem geconstrueerd. Deze grafieken zijn voor de Nederlandse omstandagheden (klimaat, bodem) toepasbaar mats de volgen de aspecten m ogenschouw worden genomen. Ten eerste zijn de geo-hydrologi sche aspecten op zeer eenvoud1ge WIJZe gemodelleerd. Daarnaast IS bij de con structte van de grafieken de afvoer van regenwater als uitgangspunt gekozen.
vi Introduction
1 . Introduction
Combined sewer overflows and hydraulic overloading of treatment plants cause quality problems within the rece1ving surface waters. The disconnection of imper VIOUS areas from the sewage system is a contribution to the solution of these problems. Disconnection is defined as "methods to reduce or deaccelerate storm water runoff from impervious or paved surfaces, w1th the purpose of reducing the 1 hydraulic load of a combined sewage system" ( 11 ) • Disconnection can be realized by guiding the storm water runoff to an infiltration and/or percolation facility.
A literature study has been carried out to examine both the current design proce dures of infiltration and percolation facilities as well as the vanous disconnection methods. lt appeared that the design procedures exhibit a number of spurious assumptions. They could lead to a false assessment of the required dimensions. The objective of this research is to remove these assumptions. Therefore a design procedure based on a realistic modelling of the infiltration and percolation process is proposed. From the reviewed disconnection methods only the following have been chosen for further research, as they seemed to be very promising for the Dutch conditions: a percolation trench with or without drain at the bottom of the facility and the German Mulden-Rigole system. This report discusses the develop ment of the new procedure and its results.
A summary of the literature study is given in chapter 2 . The shortcomings of the reviewed design procedures, the proposed improvements and the proposed dis connection methods are discussed.
As the construction of the traditional Storage-Design Discharge-Frequency curve appeared to be impossible, a new design graph has been developed. The construc t ion of this graph - based on computer simulations - is described in chapter 3 .
The computer program used for the simulations is based on a reservoir model. In chapter 4 a precise description of thrs model is given. The mass balance, the formulas used for the calculation of the inflow and outflow. and the actual schematization are discussed successively. To determine the most proper simula tion model for the desrgn, various srmulations with alternative models have been performed. The results have been compared in order to analyze the influence o f each alternative model on the required dimensions of the facility.
The relevant design graphs of a percolation trench - with various cross sections - and a percolation trench with a drain on the bottom are presented rn chapter 5 .
1 Uterature reference The Dessgn of lnftltrauon and Percolation Facslsties
The modell1ng of and the performed simulations w1th the Mulden-Rigole system are descnbed in chapter 6. Furthermore a des•gn graph of th1s disconnection technique IS presented.
Conclusions and recommendations of this study are given in chapter 7.
The tnput of the computer program consists of a 15-year-long ram senes from a research proJect of the IJsselmeerpolders Development Authonty 1n Lelystad. Two processes. loss and delay. occur during the runoff process of the ram water on its way to a fac1hty. To include these processes in the s1mulations. the ram series is transformed mto an mflow senes. The simulat1on of th1s precipitatiOn-runoff process and the actual transformation of the series are discussed 1n appendix A.
The output of the computer program consists of a series of calculated water levels in the facility. This senes is submitted to an extreme value analysis. The used statistical method and the choices made for 1ts application are descnbed in appen dix B.
AppendiX C presents a practiCal view of the application of infiltration and percola tion facilities by descnbing a case study of the city Hameln/Tundern (Germany).
Appendix D discusses the traditional design of a Mulden-R1gole system 10 the c1ty of Enschede. This design was performed w1thin the framework of a feasibility study for the application of the Mulden-Rigole (Wadi-) system m the residential area Ruwenbos.
2 Applicable disconnection methods and their design
2. Applicable disconnection methods and their design
2.1 Introduction
A literature study has been carried out in which directives and design procedures for disconnection methods of various countries have been examined. This resulted in an evaluation focussed the following topics: the various disconnection methods and the design procedures for the various facilities. This evaluation is summanzed in sections 2.2 and 2.3 of this chapter. Shortcomings of the design procedures and proposals for improvements are given in sections 2.4. Section 2.5 discussed the proposed disconnection methods.
2.2 Evaluation of the various disconnection methods
The different methods can be divided into two groups: surface infiltration and sub surface percolation facilities. All studied directives show their own specific combi nations which lead to different kinds of facilities. always based, however, on the two groups mentioned above.
A new method developed in Germany is called the Mulden-Rigole system [7) . This method is a combination of rnfiltration, percolation and drainage of ground water. The rain water infiltrates through a topsoil (with purification ability) to an underly ing gravel-filled box from which it percolates into the surrounding soil. A drainpipe is located on the bottom of the box. This pipe does not convey water into the facility, but drains the amount of water unable to percolate into the surrounding soil. This drainpipe can be integrated in the traditional drainage network of the area.
2.3 Evaluation of the design procedures
Each country has developed its own design procedure for infiltration as well as percolation facilities. Moreover, the German L6. 131 and Swiss [21 directives show design procedures for each tndividual disconnection method. The Dutch directive 11 21 contains a general procedure which is suitable for all possible disconnection methods.
The following table shows which return periods are used in the different countries for the design of disconnection facilities.
3 The Design of lnfrftratron and Percolation Facrhues
Country Oenmar~ I Germnrly I Mulden R I Holland I Swoden I Swotzorland U.K. U.S.A. T lyearl 2 I 5 1 5 I ehorce I 2 I chOice 10 2 Table 1 Return perrods used rn varrous countnes
Practically each country has developed a Storage-Desrgn Discharge-Frequency curve which IS used for the design of the disconnectiOn fac1littes. The Netherlands, Switzerland and Denmark [8] constructed these curves usrng a rarn serres, whereas the other countries constructed them using a rainfall lntensity-Durauon Frequency curve.
The runoff coefficient for calculaung the actual runoff into a conveyance system. takrng rnto account the losses that occur, IS only descrrbed in the drrect1ves of Sweden (141 and the U.S.A. (14]. In The Netherlands, when usrng the Storage Design Discharge-Frequency diagram, there is a cho1ce between curves based on rain data as well as curves based on inflow data.
The permeability of the soil is often difficult to assess. In Germany, Sweden, Switzerland and the U.S.A., this soil parameter is assumed by using tables with values for each type of soil. In Denmark and the United Kingdom [4] the perme ability is determined by measurements at the specifiC site. The Dutch procedure allows both measured and esttmated values. In Sweden, the U.S.A. and the Netherlands. measured values and values obtained by using Hazen's equation are multiplied by a safety factor.
The effective percolat1on area of the facilities ts also a parameter which is subject to great diversity. In most procedures the area is split into two different parts: 1. The bottom area of the facility. In most countries the bottom is assumed to be clogged after some time wh1ch means that the water can not percolate through the bottom to the surroundtng soil. The cloggtng is caused by small Slit and sand particals whtch enter the facility wtth the runoff water. Assuming a clogged bottom adds to the safety of the design of the facility. 2. The side area ("walls") of the facility. The effective size of this area depends in practice on the water level tn the facility. Thts level, and consequently the percolatron area, IS a tJme depend ent varrable. As most countrres design a facility usmg one desrgn storm (thus, fluxes do not vary tn umel the percolatiOn area ts a constant value. In all rev1ewed procedures the he1ght of the percolatiOn area IS assumed to be half the construction depth of the tac1lity. In none of the procedures the head ends of the facthties contnbute to the percola tion area.
4 Applicable disconnection methods and their design
2.4 Shortcomings and proposed improvements for the design procedure
The design procedures with Storage-Design Discharge-Frequency curves based on Intensity- Duration-Frequency curves exhibit the following shortcomings: 1. A point on an Intensity-Duration-Frequency curve represents a precipitation depth within a certain time interval with a certain return period. This means that effects due to consecutive storm water events and rainfall before or after the intensive part of the event are not included. The history of a rain event is neglected which could mean that the storage volume is not empty as the next event begins. This effect is especially likely to be found in systems with a slow discharge characteristic, such as infiltration and per colation facilities. lt will probably lead to an underestimation of the dimen sions of a facility. 2. The discharge from a facility is assumed to be a constant value. In reality however, this discharge varies in time. lt is determined by the percolation area which depends among others on the water level in the facility. The curves based on rain series (The Netherlands, Switzerland and Denmark) remove the first shortcoming, but the discharge from the facility is still considered as a constant value. Furthermore, the loss process is not taken into account.
T o prevent the above mentioned shortcomings, a Storage-Design Discharge-Frequency curve based on simulations with a reservoir model is pro posed. These simulations will be performed with a computer program. To consider the losses which occur during the runoff process, the input for the model is the rainfall series measured in Lelystad by a ground level raingauge ( 1970-1984). The output of the model is the water level in the storage volume, calculated every 5 minutes. With this output the required storage volume can be determined for different return periods. In order to consider the discharge as a dynamical variable, the percolation area is determined by the water level in the storage volume of the facility. This level varies in time, and consequently the percolation area becomes a dynamical variable. The hydraulic gradient is assumed equal to 1 m /m.
2.5 Proposed disconnection methods
A traditional percolation trench was selected as the first disconnection method to be modeled. Direct surface infiltration methods will not be elaborated on in this study, as they generally require a significant amount of surface area. On the other hand, water can conveniently be brought into the percolation trench by infiltration from the surface.
In many areas in the western part of The Netherlands high groundwater levels and soils with low permeability occur. These areas seem not suitable for the applica tion of a traditional percolation trench, as these conditions lead to very large
5 The Desogn of lnfoltratoon and Percolatoon Facilotoes
dimensions. Therefore such a trench could be equipped with a drain on the bot tom. The drain guarantees the discharge of water from the facility, even if the permeability of the surrounding soil is very low. The water still has the possibility to percolate into the surrounding soil. The drain is moreover meant to control the ground water to a level below the facility, so that the storage volume in and alongside the trench are available for stormwater storage. The drain discharges to surface water and can be integrated in the traditional drainage system. The trench with a drain on the bottom is proposed as the second disconnection method to be modeled.
lt 1s posstble to keep the storm water visible at the surface during the runoff process. Thts visual aspect enlarges the involvement of inhabitants in the urban water management system. Moreover stormwater pollution is visualised- and as a re sult abated- this way. The Mulden-Rigole system combines this visual aspect with the infiltration and discharge of rain water. lt is a combination of a percola tion trench with a drain and a surface storage. The drain serves the same function as in the percolation trench. The Mulden-Rigole system is proposed as the third disconnection method to be modelled.
In figure 2 cross sections of the proposed methods are presented: the percolation trench, the percolation trench w ith a drain and the Mulden-Rigole system.
Percolation trench Percolation trench with dra1n Mulden-Rigole system
...... :-::-:::- . ~:~·- -·'-· -- --c;.-
Fogure 2 : Proposed dosconnectoon methods
6 A design graph based on simulauons
3. A design graph based on simulations
3.1 Introduction
The design of a disconnection facility implies the determination of its required storage volume. As stated in chapter 2 this is commonly done using a Storage Design Discharge-Frequency curve. This curve is often based on an Intensity Duration-Frequency curve. A Storage-Design Discharge-Frequency curve based on simulations with a reservoir model has been proposed to overcome the shortcom ings mentioned in section 2.4.
However, it is evident that the construction of a traditional Storage-Design Dis charge-Frequency curve was no longer possible if the discharge from a facility is considered as a time dependent variable. Therefore a new graph had to be devel oped. In section 3.2 a summary of the construction and use of the traditional Storage-Design Discharge-Frequency curve is given. Section 3.3 deals with the development of the new design graph.
3.2 The traditional Storage-Design Discharge-Frequency curve
Use A fictitious example of a traditional Storage-Design Discharge-Frequency curve is given in figure 3.
StoraGe volume lmml 1 16
14
12
10
6
4
2
0 23456789 Discharge (m,Jts•he))
F1gure 3 : An example of a trad1t1onal Storage·Des1gn Discharge-Frequency curve
For a specific constant discharge from the facility, a point on the curve gives the required storage volume needed once per T year. T is the return period. Both the storage volume and the discharge are divided by the amount of connected imper-
7 The Design of Infiltration and Percolation Facilities
vious area (A,P lha]). This explains their units, which are respectively 1m3 /ha or mm] and [m3 /(s·ha) or mm/s].
To design a disconnection facility, first of all its dimensions have to be estimated. With these dimensions the design discharge is calculated. Next the required stor age is read from the vertical axis (dotted line in figure 3). This storage volume is compared with the estimated volume. The proper dimensions are found iterating this procedure, until the storage volume and the estimated volume are about equal.
Construction Two methods are known for the construction of a traditional Storage-Design Dis charge-Frequency curve. The first method is based on an Intensity-Duration-Fre quency curve, the second on simulations with a rain series and a reservoir model. Both methods are shortly discussed below.
Intensity-Duration-Frequenc y curve The Storage-Design Discharge-Frequency curve is constructed from the Intensity Duration-Frequency curve with the following procedure:
• A constant discharge (q 1) is drawn in the Intensity-Duration-Frequency cur ve. The maximum difference between the curve and the discharge 'line' is
the minimum required storage volume (S1 ) for that specific discharge q1 •
• The values S1 and 1 q give a point of the Storage-Design Discharge-Frequency curve.
This procedure is repeated for several constant discharges and additional storage volumes. Fitting a curve through the obtained points, a Storage-Design Discharge Frequency curve is created for a specific return period T. The procedure is shown
in figure 4 . lt is noted that the discharges q1 .•• qn are chosen arbitrarily, represent ing a certain constant discharge by percolation from the facility.
Precipitation Storage
- ~: I T Duration Discharge
Figure 4 : Construction of a SDD-curve from an IDF-curve
8 A des1gn graph based on simulations
Rain series and a reservoir model For the construction of a design graph with the use of a reservoir model a signifi cant number of simulations is required. During these simulations all model parame ters remain fixed except for one parameter. This parameter is varied for each simulation.
In case of a Storage-Design Discharge-Frequency curve, the discharge from the facility is the variable parameter. The curve is constructed from the rain series using the following procedure: • A computer program transforms the rain series into a series of storage volumes for a constant discharge (q,) from the reservoir, neglecting the effects of the rainfall runoff process. The peak values from the series of storage volumes are collected. • A statistical method (e.g. the Peak Over Threshold method) transforms these peak values into minimum required storage volumes for several re
turn periods T (points S,, 5 2, S3 in figure 5).
, Repeating this procedure for various discharges q, ... q0 the required storage volumes 5 1 ••• Sn are calculated. A curve is fitted through the obtained points to construct the Storage-Design Discharge-Frequency curve. The procedure is shown in figure 5.
Precipitation it) Storage
s, s, -- ~ Storage-- ! --- · ·- LDischarge ~ s3 '--''------~
q, Discharge
F1gure 5 : Construction of an SOD-curve from ra1n senes and a reservo1r model
For the input into the reservoir model it is better to use an inflow series instead of a rain series. In this way a more realistic construction of a Storage-Design Discharge-Frequency curve is achieved. If an inflow series is not available it is possible to transform the rain series into a simulated inflow series using a loss/delay model. The precise modelling of the described loss and delay processes as used in the reservoir model in this study is presented in appendix A.
9 The Desogn of lnfoltratoon and Percolatoon Facolotoes
3.3 Development of the new design graph
Introduction As stated in section 2.4 the construction of a Storage-Design Discharge-Fre quency curve can best be based on simulations with a reservoir model. However, if the discharge from a percolation facility is modeled as a dynamical variable, the procedure for the construction of such a curve cannot be used, since the variable parameter has to be a stationary value during a simulation. Therefore another parameter, which replaces the discharge as variable parameter, is required for the construction of the design graph.
The 'simulation' graph In order to calculate the d1scharge dynamically during a simulation, the reservoir is given fixed dimensions. To perform several simulations the length of the reservoir has been chosen as the variable parameter. The width of the reservoir remains fixed for each simulation. The height of the reservoir is very large, in order to allow for any water level. The output of a simulation run is a series of water levels in the reservoir. This series of water levels is transformed into expected water levels with different return periods with the use of extreme value analyses.
The reservoir length is scaled to the amount of connected impervious area, giving 2 a ' length-impervious area' ratio (L/Amp [m/m ]). If simulations are performed for several ratios, the calculated extreme water levels can be plotted against the length-impervious area ratio for each return period. This resulted in a 'simulation' graph as presented on the left side of figure 6.
Waterlevel (m( Waterlevel (ml
l_T 1 T2 T•
a b c L L (m/m2 j (m/m21
Fogure 6 : A 'somulatoon' graph and the determination of the reQuored domensions.
If a construction depth (dconwl for the facility is chosen, the required length-imper vious area ratio is assessed explicitly for different return periods. This is shown in
10 A design graph based on s1mulations
the graph on the right side of figure 6 by the dotted line d cons t• and the points a,b and c. As the width had already been chosen, all dimensions of the facility are determined.
Design graph The 'simulation' graph as presented in the previous section is only valid for the one specific permeability used during the simulations. By repeating the simulations for a different permeability, other graphs can be constructed. If a constant con struction depth is used, a new graph can be constructed from these various graphs. In this new design graph the permeability (K-value) is plotted against the required construction length-impervious area ratio. lt is noted that the construc tion width remains fixed for all simulations with various K-values to construct a proper graph. The construction of the new design graph is shown in figure 7.
Waterlevel [m]
a b c L [m/m' I A.... o=~s~- ~ e b ====~~ T, Waterlevel [m! d a ======~ T2 I I T3
K -~'~T, [m/si I I T2 I I I T3
d e f L [m/m2!
F1gure 7 : Construction of t he new des1gn graph from various 'simulation' graphs.
The design graph is used as follows: 1. Fo r a certain permeability and return period the required construction length-impervious area ratio is read on the y-axis of the graph. 2. As the depth and width of the facility are known, the amount of connected impervious area determines the required construction length of the facility.
it should be noted that with this design graph (or KL-graph), the iterative charac-
11 The Design of Infiltration and Percolation Facilities
ter of the design procedure with the traditional Storage-Design Discharge-Fre quency curve has disappeared.
12 The modelling of the percolation trench. calculations and results
4 . The modelling of the percolation trench, calculations and results
4 .1 Introduction
In this chapter a more precise description is given of the reservoir model used for the simulations with a percolation trench. Furthermore, four different scenarios have been set up to investigate both the sensitivity of a dynamical discharge from the facility, and the contribution of the bottom area to the amount of discharge. From a comparison between these scenarios a choice is made as to which simula tion model will be used for the construction of relevant design graphs. The results of simulations have been compared in order to verify firstly, the presumption that the effect of consecutive events influences the required storage volume; secondly, the decision to use inflow instead of rain series as input for the simulations. The final section elaborates on the effect of a drain on the bottom of the percolation trench.
4.2 The reservoir model
For the simulations with a percolation trench a computer program was written. This program is based on a reservoir model. Figure 8 shows both the reservoir model and the modelled percolation trench.
Inflow Inflow l :.. ·:':•: ;.-. ~ ..... Construction --rL __ __ ••••••... depth L Percolarr · y:~:d Storege Outflow .. ,. . \ ~ Construction width
Figure B : Reservo1r model and modelled percolatiOn trench.
mass balance The reservoir model is described by the following mass balance:
Inflow = Outflow + fj, Storage ( 1) I
13 The Design of Infiltration and Percolation Facilities
in which: 6oS change in stored volume per time step v,n amount of inflow per time step Vou t amount of outflow per time step
Calculation scheme The mass balance was translated into a numerical scheme in order to use the reservoir model for a simulation in which the water level is calculated every 5 minutes. In this numerical scheme the storage volume at a certain moment is assessed by using the storage volume calculated one time step earlier and the in and outflow during the time step. The numerical scheme is presented in formula 3. The water level in the reservoir is the output variable of the model. it is ob tained by dividing the stored volume by the length L and the width b of the reser voir (see formula 4).
s, (3)
s, h, (4) L b
in which: s, stored volume at t 1m3] s,.~~, stored volume at t - 6.t [m3]
V lit. on amount of inflow during At [m3] VA t, out amount of discharge during 6.t 1m3] h, water level at t [m] L length of the reservoir [m] b = width of the reservoir [m] 6.t time step Is]
The initial values of the storage volume and the water level are 0 m3 and 0 m respectively.
Inflow The amount of inflow during a time step is read from the inflow series. The series also give the following information: the date, starting time and duration of the inflow event. The inflow figure is multiplied by the amount of connected impervi ous area to obtain the inflow volume. The length of the time interval between two inflow events is unknown. By using the known ending time and date of the first event and the starting time and date of the second event, this period is calculated. In this way a continuous series is created. During the time interval between two events the amount of inflow is 0 m3 /s for every time step.
14 The modelling of the percolation trench, calculations and results
Outflow The amount of percolation outflow during a time step equals the percolation rate multiplied by the duration of the time step. The percolation rate is described with the help of Darcy's Law:
Opere = K I Aperc (5)
in which: Q pe,c percolation rate [m 3/sl K permeability [m/sl hydraulic gradient [m/m] Ape•c percolation area [m2]
lt should be noted that Darcy' s law describes the ground water flow in saturated soil. The hydraulic gradient is assumed to be 1 m/m. If the soil is unsaturated the hydraulic gradient can be significantly larger due to the suction stresses of the soil matrix. Consequently the assumption of 1 m/m will lead to a safe design. On the other hand, the permeability K may be reduced in unsaturated conditions.
For the size of the percolation area different assumptions can be made. In section 4.3 these assumptions will be elaborated on. The drain discharge is modeled as a constant value per time step. This discharge is added to the percolation discharge, to obtain the total outflow per time step.
4.3 Four different scenarios for the percolation rate from the reservoir
One of the objectives of this project is to investigate to which extent a dynamical percolation discharge will influence the amount of required storage volume com pared to a stationary discharge. Secondly, the contribution of the bottom area to the discharge from the facility and its influence on the required storage volume is examined. Therefore four scenarios have been set up, from which the results are compared.
The percolation area of a percolation trench is defined as: "The area through which water from the storage volume percolates into the surrounding soil". In general, this area can be divided into three parts: 1. The head ends: The head ends will hardly contribute to the total amount of the percolation area if the length of a facility is much larger than the width (L,. b). As this is very likely these areas will be neglected. 2. The bottom:
15 The Des1gn of lnf1ltrauon and Percolat1on Facilities
The bottom area of a facility will contribute to the percolation area in case of an unclogged bottom. 3. The side walls: In practice the active area depends on the water level in the facility. If the water level is calculated dynamically, the effective area becomes time dependent as well. Consequently, the percolation is to be calculated for each time step. However, in several of the current design procedures the side area is considered to be constant in time. The height of this area is then often assumed to be half the construction depth of the facility. With this assumption the discharge remains a stationary parameter.
The three different parts are shown in figure 9 below.
Side area: Bottom area: - ' ::,... ' . -,' ... ~ ...
Head end:
F1 gure 9 : The three different parts of the percolation area
Now the four scenarios concerning the total percolation area can be defined:
1. Clogged bottom and a stationary side area:
A L · 2_d = L d perc , constant = 2 2
2. Unclogged bottom and a stationary side area :
A perc , constant = Ld + Lb
/ 3. Clogged bottom and a dynamical side area:
16 The modell;ng of the percolation t rench, calculations and results
1 A perc , dynam1c. = 2Lhw
4. Unclogged bottom and a dynamical side area:
1 A perc , dynamJc.= 2Lhw + Lb
In the next section the results of the simulations with these scenarios are dis cussed.
4.4 Simulations with the four scenarios
KL-graphs as introduced in chapter 3 are used to compare the results of the four scenarios. All simulations were executed with the same trench dimensions: a construction width of 0.5 metre and a construction depth of 1.0 metre. The facility is filled with gravel giving a porosity of 0.35 for the storage volume. The graph is constructed for the following return periods: 2, 5 and 10 years. The results of two scenarios are shown in the same graph.
A stationary discharge vs. a dynamical discharge For this case the scenarios 1 and 3 are compared. Figure 10 shows the results.
Percolation trench Construction depth = 1 .0 m, width = 0.5 m
T=10
T=S
T = 2
: l ; i. ·-·~· ·------~~1l1i· r-·---r--·· . \ !!i
t ! t t !!i i , I !! ' I i ! ! ! i! ! ! ! j j1 ; '; i li; . ; ; ! 0 ~--L-~~LLLUL---~-L-L~~~--~--~-LLLLU 1.0E.Q7 1.0E.06 1 .0E.QS l .OE-04 K-value (m /s}
Figure 1 0 : Stationary discharge vs. dynamical discharge
From the figure it can be concluded that for a low permeability there is a differ-
17 The Design of Infiltration and Percolation Facilities
ence in required construction lengths between the two scenarios. The scenario with the dynamical discharge results in longer facilities. For soils with a better per meability the required dimensions do not differ significantly. In case of low perme ability the difference is probably caused by the fact that the facility with a dynam ical percolation area empties more slowly. Low water levels in the facility give a small percolation area. This contrasts with the stationary scenario in which the height of the percolation area is always half the construction depth. The probabil ity that the storage volume is not empty before the next event starts is therefore larger in scenario 3. For the construction of relevant design graphs a model with a dynamical percolation area is to be recommended.
Clogged bottom vs. unclogged bottom For this case the scenarios 3 and 4 are compared. Figure 11 shows the results.
Percolation trench Construction depth = 1.0 m, width = 0.5 m
0 '3 r-----~~--~~--~~--~1 -j-,j~:~, ------i~i~!r;l <'1 ' ; ;; < T=10 E ] T = 5 ~ 0.2 .. T=2 "'i:l - Clogged bottom ·;; ··· Unclogged bottom Q; c. .E 0.1
: . ; : i ~ : i ;: J1 . . li l' :: 0 L---L-~~~~--~~~~WU~--~~~~~ 1.0E-07 1.0E-06 1.0E-05 1.0E-04 K-value (m/s)
Figure 1 1 : Clogged bottom vs. unclogged bottom.
From the figure it can be concluded that there is a significant difference between the two scenarios. For each permeability the scenario with a clogged bottom requires larger construction lengths as the bottom area has a substantial share in the discharge from the facility. For the construction of relevant design graphs application of a clogged bottom scenario is recommended, as the bottom of the facility can get clogged after some time.
18 The modelling of the percolation trench, calculations and results
4.5 Simulations with other scenarios
In this section some other scenarios are elaborated on. First the sens1t1v1ty of a percolation trench for consecutive inflow events is discussed. Secondly a compari son between the results of simulations with rain data and inflow data is made. The reservoir used in this simulations is the same as described in section 4.4. As re commended, the percolation rate and bottom conditions are chosen conform scenario 3.
Consecutive events vs. non-consecutive events To include the effect of consecutive events a reservoir model with a continue series is used for the construction of the design graphs. In the non-consecutive scenario the storage volume is considered to be empty at the beginning of every new inflow event.
Percolation trench
;;; T = 10 ~ _g T=S ~ 0.2 --.-..f ' T=2 ~ " - Consecutive ·;;:5 ···· Non-consecutive li c. .§ 0.1
! !
0 L---~~~~~~--~~~~~~--~~~-L~LU 1 .O E-07 1.0E·06 1.0E· 05 1.0E·04 K-value (m/si
Figure 12 : Consecutove events vs. non-consecutive event s.
From the graph it can be concluded that the effect of consecutive events is of great influence on the required dimensions of the facilities. Although the effect is limited in soils with a high permeability, both scenarios show big divergence in soils with low permeability. The choice for the use of a reservoir model with a continuous series is therefore justified.
Rainfall vs. inflow The input for the rainfall scenario is the original rain series from Lelystad. The second scenario uses inflow data as input. The model to transform rain data into
19 The Design of Infiltration and Percolation Facilities
inflow data is discussed in Appendix A.
Percolation trench Construction depth = 1.0 m, w idth = 0.5 m
0.3
i 111. i i 1 J l j i i ! l; c::< I J! ;, ~ ! I i ! ! f 1 J I T = 5 I ; I i l I ll i _; i i i i :. ; : ______._. __ ~J..JJ. ~ 0.2 ·------·-~ --t---~-:- - f - i-t~ - t; J llli I i I; i ~ l ! I !; i ·;;~ i ~ ll I '! : t; Q. i l !l .S o.1 1-··+ -i-+-+i ··· ~ ... ;. ... : : l : £ i j i. "'ti ...J :! Ii '''i I j 0 L---~~~-L~LW----~-L-L~~~----'~~~LL~U 1 .OE.Q7 1.0E.Q6 1 .OE.Q5 1 .OE.Q4 K-value (m/si
Figure 13 : Ramfall vs. Inflow.
lt is clear that the scenari o w ith inflow data results in smaller required dimensions for the percolation trench. As the inflow data represent a more realistic view of the amount of water entering a facility, they w ill be used for the construction of the relevant design graphs.
4.6 A percolation trench with a drain on the bottom
As stated in section 2.4 the percolation trench can also be equipped w ith a sub surface drain on the bottom of the facility (see figure 2). In areas w ith low perme ability it is expected that the contribution of the drain to the discharge from the facility will be significant. The drain starts discharging if the water level in the facility rises above 10 cm. The discharge capacity is assumed 2 1/s/ha. The dimen sions of the drain are neglected. The trench is modeled conform scenari o 3.
20 The m odelling of the percolation trench, calculations and results
Percolation trench Construction depth= 1.0 m, width = 0.5 m, q = 2 1/s/ha
0.3 .------,-,---,-,-----~-..,.--.,.., .,.., .,.., :-, ----~-.,..,..,., .,.. ; .,..,;
T ~ 10 c;; ( .E T =5 _g T=2 .. 0.2 .,~ .. T= 10 ~ 0 -~ T=5 "a. .§ 0 .1
.J:: 2> _J"
0 L--~~~~UU~-~-~~-JW~~--JL~~-LLLLU 1.0E- Figure 14 : Dram vs. no drain From the figure it can be concluded that there is a significant difference in re quired dimensions between the two scenarios. Especially in areas with low perme ability the influence of the drain almost halves the required dimensions even though the discharge of the drain is relatively small. Therefore it is recommended to consider this type of facilities in areas with low permeability and high ground water tables. The design graphs of the percolation trench with and without drain are presented in chapter 5. 2 1 The Design of Infiltration and Percolatoon Facilities 22 Design graphs percolatoon trench 5. Design graphs percolation trench In this chapter a number of rele vant design graphs for a percolation trench and a percolation trench with a sub-surface drain on the bottom are presented. The reservoir used to perform the simulations for the construction of the graphs, is modeled according to the conclusions from chapter 4: The percolation area is considered as a dynamical variable and the bottom is assumed to be clogged. The input for the simulations is a continuous series of consecutive inflow events. This inflow series is obtained from the Lelystad rainfall data with the assumption that for every inflow event an initial loss of 1 mm occurs. No losses occur if an inflow event, due to the delay process, overlaps the next rain event. The construction depth has been chosen at 1 m . The minimum construction width of the facilities has been set at 0. 5 metre. The porosity of the trench filling mate rial is 0.35. lt is assumed that the trenches will be excavated mechanically. There fore the width of the shovel of the excavator is decisive. Design graphs with widths of 0.5, 0.75, 1.0, 1.25 and 1.5 m for the percolation trench are presented. The range of K-values varies from 5 * 1 o·5 to 5 * 107 m/s. This range reflects the permeability of frequently occurring types of soils in the Netherlands. A design graph shows three curves, representing the return periods 2, 5 and 10 years. The decision which return period should be used, is left to the user of the graph as only he can estimate the risk in case a facility fails. The inflow series contains 1 5 years of data. This is a short period to give reliable design dimensions for return periods larger than 5 years. The discharge of the drain on the bottom of a facility is 2 1/(s·ha) connected impervious area ( = 0. 72 mm/hour). This value reflects the design discharge from a rural area. Furthermore graphs have been constructed with 5.6 1/(s·ha) ( = 2 mm/hour) and 14 1/(s·ha) ( = 5 mm/hour) as drain discharge. it should be noted that the design graphs show the exact results from the simula tions. Smoothing of the curves is often done to make the results look more coher ent. As there is neither a statistical nor a physical reason to smooth the outcomes of the calculations, there is no scientific reason for smoothing the curves in the graphs. The simulations with wide facilities showed some statistical disturbances, espe cially for low permeability. Although wide facilities have the same percolation rate as narrow facilities, the amount of water stored at a certain water level is larger. Consequently the emptying time of wide facilities increases compared to narrow f acilities. Simulations with wide facilities resulted in the reservoir being empty 23 The Design of Infiltration and Percolation Facilities only 3 to 4 times in a period of 15 years. If this is the case, the condition for a reliable statistical analyses is not met. Therefore the minimum number of calcu lated peaks is set to 40. Only simulations meeting this condition have been used for the construction of the graphs. In the design graphs for the percolation trench without a drain, the ratio between the construction length (L) and the construction depth (d cons~< ) is linear, as the simulated outflow only depends on the product of L and the water level in the trench. Also a sensitivity analysis proved that for the determination of the required dimensions it is allowed to variate the required d consto and L as long as their product remains equal. Consequently the y-axis of the design graph can be converted into . (L *dconst•)/ A.,0 This offers the user of the graph the possibility to fit the facility more flexible in the urban infrastructure. However, it is advised to variate the construction depth only if necessary and to keep the construction depth within the range of 0.5 to 1.5 metre. The bottom of the facility should in neither case be situated below the ground water level. 24 -. Percolation trench Construction depth = 1 .0 m, w idth 0.5 m 0 .3 N < E E ro 0.2 .....(!) ro +T 2 (/) ::J 0 ·;;; - ;I:,,: + - rr-III!t -+-T 5 ..... -~ -~~ r rr~lrrr ~ - (!) a. +-T 10 E 0.1 ..._ ..c.... 1 1 cC> 11 Jr- I t i + , ,, (!) r[ ]l l ...J 0 1.0E-0 7 1.0E-06 1.0E-05 1.0E-04 K-value (m/s) Figure 15: Percolation trench, construction depth = 1.0 metre. width = 0.5 metre 0 N I!) 1- 1- 1- + • 1 <;to 0w 0 ~ ~ =~~ ~ ~ ~ ~ ~~~ ~~= ~~ -=~ ~I ~~ ~~~~~ ································r·······························i····························· ...... ,:...... 1 ...... ;: ..... ···················· E ~:~:::r-~ :r: ::r :: : I!),..... I!) 0 0 I 11 ...... i······"'"''''''"'''''''''''''l''''''''''' . w 0 ..c. ::::::::::::::::::::::::::::::::r:::::::::::::::::::::::::::::::r:::::::::· .. :::::·.::::::·· ..:T:::::::::::::::::::::::::::: £...... u "0 c - ~ Q) ·:::_:·::··:::-::::.:.:::::::::r:::·::·::::::.: .. :::·::_::·::::-:::::::··::::·::::::::-:::::::r::::::::::::::::::::::-:::: Cl) ...... E +-' ...... L...... -:•••• ...... ·········••••-:••·························· E c 0 -- ...... Q) 0 ...... ,...... ··················~····························· ~ ·- :J a; ..... 11 ro ro > E ..c. I Ln 0 ..... ~ u a. 0" ...... Q) .c Q) "0 -- r --: - ·r-- - c... c CD 0 ""~ ..... 0 u ...... ·····!································!····························· w ~ :J ...... ··!· ...... ~ ...... ~ ...... q a; ...... E Cl) c ::::::::::::::::::::::::::::::::.::::::::::::::::::·:::::::::::::~::::::::::::::::::::::::::::::::.::::::::::::::::::::::::::::: ~ 0 .c u ...... 1 ...... 1...... 1 ...... a. Q) "0 c 0 ~ su "'c -:: :r-: 0 ~~~-::::::~~~:::: u .cu c ,..... ~ 0 c 0 w -~ N I!) I!) oq 0"' u 0 . 0 q Q; 0 0 a. CD (lvW/ W) eaJe snO!/\Jadwl 1 l.Jl6Ual Q) 5 u:Ol Percolation trench Construction dept h = 1.0 m, widt h 1.0 m 0.2 N < E ]: 0 .15 Cll ....QJ 1 ' ' tttt t t ' t iiTIT' Cll l!!li ; ' •T 2 Cl) :J .Q 0 .1 - ~ -+-T 5 2: 1 1 QJ -.k-T 10 0. . , 11 E I I , ! i , .c-- 0 .05 ...... ''"''''''i······ ······· !""····!· ·····t ·····i····+ .. ·t··+··+ .. ·····················t·"''''"' ···i···· '"''!'''"' 0, c QJ .....1 I I 1111111 I I I ; i ! 0 1.0E-07 1.0E-06 1.0E-0 5 1.0E-04 K-value (m/s) Figure 1 7 : Percolation trench, construction depth = 1 .0 metre, width = 1 .0 metre -- Percolation trench Construction depth = 1.0 m, width 1.25 m 0 .2 N < E 'E o .16 ro ....Cl> ro r rrlrrrrr----r ! r trrrrr ..... r 2 (/) ~-rr1rrtr ::J .2 0 .1 -+-T 5 >.... a.Cl> 1 +-T 10 E L-llrlll i i -11111! ..._ 1-li-lllll i ..c. 0 .05 ···················•·············•·········•·······1·····•····1···•···•···•·······················•·············1········· ...... ,... •.. El c: Cl> _J i I i I i I i 11 ! I , , ! !! I 0 1.0E-07 1.0E-06 1.0E-05 1.0E-04 K-value (m/s) Figure 18 : Percolation trench, construction depth = 1.0 metre, w1dth = 1.25 metre Percolation trench Construction depth = 1.0 m, width = 1.5 m 0.15 N" < E .s-- ro 0.1 ....Q) ro -e-T= 2 (/l ·- -rrrrrrtr·t ! i 1·rr1rrr ::::l ll!llil 0 1 j 1 i 1 i j j j : 1 1 i j i ! l l ! ·::; -+- T = 5 .... Q) ....._ T = 10 0. 11111·1 I ; 11 1 1 11 I I I E - 0.05 ·····················• ············ .•...... l ...... l ..... l ....i .... i... l .. l ...... i...... •.. ~ ...... i ...... !...... i...... i...... i. ... l ... l ... ;.. --..c..... C>c Q) ...J i Ill! IIll l ' l ' : I i i 11 ' 0 1.0E-07 1.0E-06 1.0E-05 1.0E-04 K-value (m/s) F1gure 19 : PercolatiOn trench, construction depth = 1.0 metre. w1dth = 1.5 metre Percolation trench with drain Construction depth = 1.0 m, width = 0.5 m , q 2 1/s / ha 0.2 N < E --E 1 ~ j ~ -; 0.1 5 •• • • • 0 ...... 0 0. j ••••••• 0 •• ~ •••••• ·i· ....; .... ~ .. 0 ; •• ~- ,j .. j ...... •.... 0 ~ ••••••••• ,j ...... ~ ..... j ..•. ; .• -~·. i. -~' .....Cl> CO 11 ! 11 I t/) I ; : i : i i ll ! !l' :::::1 •T=2 0 kl ! '!!! ] : ::: !!!! : : : : : ::: '>..... 0.1 •T=5 Cl> ·--···········r·----·--r···--r--TTTlT c. E +T = 10 --£ 0 .05 0> c Cl> -l 0 1 .OE-07 1 .OE-06 1 .OE-05 1 .OE-04 K-value (m/s) Figure 20 : Percolation trench w1th sub-surface drain on the bottom, construction depth = 1.0 m, w1dth = 0.5 m and dra1n discharge = 2 1/s/ha Percolation trench with drain Const ruction dept h = 1 .0 m , width = 0 .5 m , q = 5 .6 1/s /ha 0 .2 N < E --E 0.15 + mm ctl r ...... rr r r·· Q,) rr ...... r rr ..... ctl (/) :::l ...-T=2 0 ···············t········· ...... ··· · ······~···· ·· ·i·····t · ··t·· >..... 0 . 1 ··t·· +T = 5 Q,) c. E 1 +T = 10 ! I i , ! l I ' --L li 15, 0 .05 ...... ~ ...... ; ...... ; .. ~ ...... ;...... ~...... i ...... j ..... ,o. ••• i .... J, .. . c Q,) -1 j [ Ji I i 1 I . i ' 0 1 .OE-07 1 .OE -06 1 .OE-05 1 .OE -0 4 K-v alue (m/s ) Figure 21 : Percolatoon trench with sub-surface drain on the bottom, construction depth = 1.0 m, wodth = 0.5 m and drain discharge = 5.6 1/s/ha Percolation trench with drain (heigth = 1 .0 m, width = 0.5 m, q = 14 1/s/ha) 0.2 N < 1 ..._E ______1 1 s 0.15 .•••••••••••••..••.••.•.•. ~ ••••.• 4- ••••• j •••• j •••.o. .. j •• ~.,o. •.•...... •..... j •••••••••• .o, •••••• ,o. •••• .o. .... j ...j •.• j •• j •• j ...... o. .•.••••.•. j •••.••• j ••••• ,o. ••• e, •.• j •. .; •. j •. CO Q)..... 1 CO 1 1 (/) ' i I ' i ! I! I ' i i 1 ! 1 : ' ! I i 1! I :::J •T=2 0 > 0.1 ..... -----··········r-······r-·l····:····i··-r-l-n··················i··"''''l'''''t'"'l"l'l'lll''''''"''''''''l"'''''''l''''''l''''l"l"lll' +T=5 Q) 0...... T = 10 E ! i i :: i ..._ I I I · : :11; ! i . l ! I Ill: I J:: ~ 0.05 c Q) -jJJ il\inmmmj .. •Jli i MlT~(,1f !+i ...J I I I ! 11111 I I i i ii Ii i i i i : i i!I 0 1 .OE-07 1 .OE-06 1 .OE-05 1 .OE-04 K-value {m/s) Figure 22 : Percolation trench with sub-surface drain on the bottom, construction depth = 1.0 m, w idth = 0.5 m and dram discharge = 14 1/s/ha The Mulden-R1g ole system 6. The Mulden-Rigole system 6.1 Introduction The German Mulden-Rigole system has been developed to combine the benefits of both infiltration and conventional drainage. The result is an infiltration system with the charactenstics of a drainage system. Depending on the permeability of the surrounding soil the water can either be infiltrated or discharged v1a the drain. In the ultimate case, the design even accounts for surface drainage via overflows in the Mulden. Moreover, the system is less sensitive for high groundwater levels due to the application of a drain on the bottom of the facility. 6.2 The model A sketch of the Mulden-Rigole system is given in figure 2 (section 2.5). The Mulden and the Rigole have been modeled as two separate reservoirs. The outflow of the first reservoir ( =Mulden) is the inflow of the second reservoir ( =Rigole). The Mulden has a triangular shape. Two discharges from the Mulden can be defined: infiltration through the topsoil, and discharge through the 'slokop' if the water level reaches the 'slokop level'. The 'slokop' is the direct connection be tween the Mulden and the Rigole. The amount of infiltration depends on the infiltration area. This area depends on the water level in the Mulden and the infiltration capacity of the topsoil. The discharge from the slokop depends on the size of the slokop and the permeability of the gravel with which the slokop is filled. The height of the Mulden-reservoir is considered as infinite to allow for any water level. A sketch of the modelling of the Mulden is given in figure 23. Mu/den Slokop htHght Slokop Figure 23 : Model of the Mulden In which: 0 1 = Infiltration from the Mulden 0 2 Discharge through the slokop tana = (depth Mulden) I ( Y. width Mulden) 33 The Des1gn of Infiltration and Percolation Facilities The Rigole is modelled as a percolation trench with a drain on the bottom. The percolation characteristics equal the characteristics of scenario 4 (section 4.4). The bottom is assumed to be unclogged as very little sediment will reach the Rigole due to the fact that the topsoil traps all the sediment. Only via the slokop small amounts of unfiltered water can enter the Rigole Rigole ·••:'•::.·• • •• .···•:'•·-·· .. ••• .···-~.&.----~~=- a, F1gure 24 : M odel of the Rigole In which: Outflow from Mulden ( = Infiltration + Slokop) Percolation from the Rigole Drain discharge 6.3 The optimal use of the system From the design of the Ruwenbos Mulden-Rigole ( = Wadi) system (Appendix DJ it appeared that during the design process neither the storage volume of the Rigole nor the drain discharge have been considered [5). However, simulations with different scenarios in section 4.6 showed that the application of a drain has significant influence on the required storage volume. Therefore both the drain discharge and the storage volume of the Rigole have been added to the simula tion model of the Mulden-Rigole system. Moreover, the slokop is used as an overflow device for rain events with a return period larger than 2 years. Consequently, during events with a return period small er than 2 years, the storage capacity of the Rigole is not used. If this storage volume is used more efficiently, the required dimensions might decrease. There fore several simulations have been performed with different slokop heights and slokop diameters. The starting values for the simulations of both the parameters and the dimensions 34 The Mulden-Rigole system of the Mulden-Rigole element have been taken from the design of the Ruwenbos system [5). They are listed below: Mulden : Width = 3.0 (m] Depth = 0.35 [m] 5 K ,nfrltrat10n = 5.0 * 10" [m/si Slokop: Slokop.,•• = 0.008 [m 2/ml Kslokop = 1.2 * 10·3 [m/si Slokop"••gth = 0.25 [m] Rigole: Width = 0.5 [m] Construction depth = 0.6 [m) Porosity = 0.35 [-] Drain: Discharge = 2 (1/s/ha] From the simulations can be concluded that a change in slokop dimensions does not influence the required dimensions significantly. The lowering of the slokop level results in a more frequent use of the storage volume of the Rigole during small rain events. However, during heavy rainfall the slokop starts discharging as the Rigole is already filled for the greater part; no storage volume is available. The most striking conclusion of the simulations with the model of the Mulden Rigole system is that the required construction lengths do not depend significantly on the permeability of the surrounding soil. This was already noticeable in the design graphs for the percolation trench with a drain (chapter 5}. If a surface storage (Mulden} is added to the trench (Rigole}, this effect is even more clearer. This surface storage (porosity = 1} has five times more storage capacity than the sub-surface storage (porosity = 0.35} and is therefore decisive. The occurring peak water levels in the Mulden reservoir are used for the deter mination of the required dimensions of the facility. During the simulations the outflow characteristic of the Mulden does not change as the permeability of the top soil is a constant value. Consequently, the required dimensions should not change either. The Rigole reservoir is loaded uniformly as the peak loads - caused by rainfall - on the Mulden are flattened by the top soil. The permeability of the subsoil, surrounding the Rigole, does vary during the simulations. This variation of the permeability can only influence the required dimensions if the water level in the Mulden is significantly affected by the water level in the Rigole. The conclu sion that - for the chosen parameters - the permeability does not significantly influence the required dimensions could be explained by the aspects stated above. 6.4 The design graphs of the Mulden-Rigole system In this section two graphs are presented for the design of the MR-system. The first graph is based on a Mulden with a width of 3 metre, the second one on a 35 The Desogn of lntoltratoon and Percolation Facilities Mulden with a width of 2 metre. The other dimensions have been chosen as presented in section 6.3. The drain discharge is 2 1/(s·ha) connected area. Mulden-Rigole element (Width Mulden = 3 meter) 0.05 .----~-~~,~,~, ~, ,~,----.,.....-~...,..,-- -~-~~.,..,..,., i ! l : ! ' . . ~ : ! ! 1 11 ~ \! H ! : ;; ...... ···+·-L.l... l . ..i .. l .. Lf ...... ~------.1 ·······•···~•·-' .;----·r---= .....,i- .. ;... i 0.04 ! :; j ! ! ! ~ ! 1 ; . ; i; i; j! c :;D 0.03 . T;2 1 ~ ~ l i ~ -+ T;5 ··-·...,.·.._,;4t>-- -.._...... --+' ~·H-' ~:.''-': ··· --i'- Q L--~-L~~LU~-~~-L~~~-~~~-LLLLU 1.0E-07 1.0E-06 1.0 E- 05 1 .OE·04 K-value (m/si Fogure 25 : MR-element with w idth Mulden = 3 metre. Mulden-Rigole element (Width Mulden = 2 meter) 0 ·05 ,...--.,.--...,...-,...,...,.,..., '"'. ,"".- --,-.,.-,.....,....,,_..... _i""'".i""i .:- - .,.--.--,-,.-_i.,...i '"".i ,.....,_ '· . : !;! ..§. 0.04 :: l ~ i \ i :::: '. .. : ;l "c >----<'-+-+---'-+-+04H+----'-+-t--7 ....' : i i i 0.03 ···-····i·-···--·1 ...... l .. -i ... f H·H···-········-t········t-···+··+·t·t··H· •T ; 2 ~ !i ·;; : :i + T;5 ~ . . . : ~ !! ~ 0.02 ...... iti~r *T ; 10 i! ! ! i !l : l :: -£ i:: :; ! ! ~ 0.01 ····-l-----)····-~···i···~ ' . ... \ .... _l_.·.~... i.. '.'··'·'· ...J ·····[·· ··-+ :! i i i l ~ ; ij 0 L---~-L~~UU~--~~-L~~U---~~~-LUWUJ 1 .0E-07 1.0E-06 1.0E·OS 1.0E-04 K-value (m/s) Figure 26 : MR-element woth width Mulden = 2 metre. 36 D1scuss1on, Conclus1ons and Recommendations 7. Discussion, Conclusions and Recommendations 7. 1 Discussion In order to design a disconnection facility in a realistic way, a new, easy-to-use K L design graph has been developed. The graphs are developed from a 'discharge' point of view. The purpose of the disconnection facilities is to discharge the rainwater by means of both percolation and drainage. The design graphs therefore present minimum required dimensions. However, the point of view from which a facility is designed influences this design. If, for example, the objective is to detain a maximum amount of rainwater in its own area, the ground water recharge is to be maximized. This can be done by raising the level of the drain at the bottom of the facility in combination with a large storage volume in a more flat (shallow) facility. The K-L graphs do not give any information about the constructive as pects of the facilities. For this one is referred to specialised literature. Furthermore, it should be noted that tor the construction of the K-L graphs the gee-hydrological aspects were simulated in a simplified way. The vari ation of the ground water flow outside the f acility was ignored by assuming the hydraulic g radient equal to 1 m/m t or each simulation. In practice however, this hydraulic g radient varies in time and therefore influences the design of the facility. More over, the ground water (level) was ignored for each si mulation. However, in practice the ground water level varies in time and could possibly influence the outflow characteri stics- and consequently the design- of a percolation facility. The K-L graphs have been construct ed using a 15 year-long rain series. From a statistical point of view o ne should realize that, when using the K-L graph, the value for a design for a return period of 10 years is less reliable than a design tor a return period of 2 or 5 years. Finally it should be noted t hat for the design and application of the proposed disconnection facilities also a number of other aspects should be taken into ac count. For instance: the quality of rainwater runoff, the integration of infiltration and percolation facilities in the urban water system, the influence of disconnection facilities on the traditional sewer system. For a proper design an integrated ap proach is needed w hich takes all aspects into consideration simultaniously. If one decides to use the presented K-L graphs he or she should be aware of the aspects mentioned above. 37 The Des1gn of Infiltration and Percolation Facilities 7.2 Conclusions The most important conclusions of this study are: The objective of this project was to develop a procedure with which disconnection facilities can be designed in a realistic way. This resulted in new, easy-to-use K-L graphs. The construction of traditional design graphs exhibited some shortcom ings. For the construction of the K-L graphs these shortcomings have been re moved. One of the shortcomings of the traditional design procedures is the assumption to consider the discharge from a percolation trench as a stationary variable. A com parison between a simulation with a stationary discharge and a dynamical dis charge showed that a simulation with a dynamical discharge requires slightly larger dimensions. This is even more noticeable for soils with low permeability. The application of a drain on the bottom of a percolation trench strongly reduces the required dimensions as compared to a trench without a drain. Moreover an increase of the drain capacity results in a decrease of the influence of the per meability on the required dimensions. Considering the German Mulden-Rigole system, the K-L graph showed that the permeability has no longer a significant influence on the required dimensions. lt can be concluded that the application of percolation facilities with a drain and the Mulden-Rigole system is very well possible in soils with low permeability. Besides, the drain enables both mentioned facilities to function properly in areas with high groundwater levels. 7.3 Recommendations For further research the following is recommended: lt is suggested to investigate the possible construction of a design graph for faci lities which are intended to maximise the ground water recharge. In such a graph the design parameters could be the desired percentage of the total amount of rainwater, used for ground water recharge. lt is recommended to investigate the geo-hydrological aspects of the infiltration of rain water with the use of a 20 simulation. The influence of the hydraulic gradient on the percolation rate, and the influence of infiltration of rain water on the ground water level should be the subjects of a next research project. 38 Doscussoon, Conclusoons and Recommendations Application of percolation facilities with a drain implies that the surplus of rain water is discharged to the surface water. Traditional subsurface drainage systems applied 1n urban areas have the purpose to control the ground water level. it is advised to investigate the consequences of the integration of these two functions into one subsurface drainage system for the design and the operation of the total drainage system. In this project simulations have been performed with a rain series measured over a period of 15 years. In order to obtain statistic reliable results for return periods larger than 5 years, it is recommended to use a longer representive series of rain data. The exact consequences of an inundation of a disconnection facility have not been analyzed in this project. it is suggested to examine these consequences and also to look into the question which return period should be recommended for the design. 39 The Des1gn of Infilt rat ion and PercolatiOn Fac1lit1es 40 References References 1. Adams R., Albert G., Grotehusmann D. (1992), Stadtebauliche Bedingungen fUr ein umweltvertragliches Entwasserungskonzept- Fallstudie Hamelnl Tundern, Zeitschrift Stadtentwasserung und Gewasserschutz Heft 18, Germany. 2. Amt fur Gewasserschutz und Wasserbau (1991), Retention und Ver sickerung von Meteorwasser im Liegenschaftsbereich, Planungsgrundlagen und Beispiele, Baudirektion des Kantons Zurich, Switzerland. 3. Buishand T.A., Velds C.A. (1980), Neerslag en verdamping, Koninklijk Nederlands Meteorologisch lnstituut, De Silt, The Netherlands. 4. Digest 365 (1991), U.K. Design Procedures, Building Research Establish ment, Garston, Watford, England. 5. Geldof G.D., Braal A.J. de (1994), Waterhuishouding plan Ruwenbos - Haalbaarheidsstudie wadi-systeem, Tauw Civiel en Bouw bv, Deventer, The Netherlands. 6. Gesellschaft zur Forderung der Abwassertechnik eV. ( 1990), Construction and Dimensioning of Facilities for Decentralized Percolation of Non-Harmful Polluted Precipitation Water, ATV Standard A 138, Germany. 7. Grotehusmann, D. & F. Sieker et al ( 1992), Naturnahe Regenwasser ent sorgung durch Mulden-Rigolen-Systeme, Korrespondenz Abwasser, 39. Jahrgang (blz. 666 tl m 687), Germany. 8. IDA Spildevandskomiteen (1995), Nedvsivning af regnvand-dimensionering, Skrift nr. 25, lngeni0rforeningen i Danmark, Denmark. 9. NWRW Report part 4.3 (1989), Neerslag, inloop, overstortmodel; Beschrijving en analyse, 's Gravenhage, The Netherlands. 10. NWRW Report part 7.1 (1986), Verhard oppervlak en watervervui/ing; Literatuurstudie en inventarisatie van ervaringen, 's Gravenhage, The Neth erlands. 11. NWRW Report part 7 .2.2 (1990), Afkoppeltechnieken voor verharde opper vlakken, 's Gravenhage, The Netherlands. 12. NWRW Report part 7.2.3 (1990), Leidraad afkoppelen van verharde op- 41 The Des1gn of Infiltration and Percolation Facilities pervlakken, 's Gravenhage, The Netherlands. 13. Sieker, F & R.W. Harms (1987), Entwasserungstechnische Versickerung van Regenwasserabflussen, Aus der Arbeit der Abwassertechnischen Vereinigung e. V., Germany. 14. Urbonas, B. & P. Stahre (1993), Stormwater, Best management practices and detention, PTA Prentice Hall, New Jersey, USA 15. Ven, F.H.M. van de (1989), Van neerslag tot rioolinloop in vlak gebied, Ministerie van Verkeer en Waterstaat, Rijkswaterstaat, directie Flevoland, Lelystad, The Netherlands. 42 Appendix A. Simulation of the precipitation-runoff process Appendix A. Simulation of the precipitation-runoff process A.1 Introduction The rain series used in this research originates from the continuous precipitation measurements carried out in different areas of Lelystad - amongst others a hous ing estate and a parking lot using a Recover ground-level raingauge. This series contains 5-minutes precipitation depths for the period of 01-01-1970 to 31-12- 1984. For a precise description of the measuring program and the characteristics of the measure areas one is referred to Van de V en I 15). Besides, inflow into the sewer was assessed in order to examine the difference between the amount of rainfall in an area and the amount of water which actually enters a sewer ( = inflow). Two processes, losses and delay, occur during runoff of precipitation on paved surfaces. The loss of precipitation is defined as all water not discharged to a facility. The delay process can be described as a transformation and flattening of the precipita tion wave on its way to a facility. Both losses and delay influence the amount of inflow and consequently the determination of the required storage volume. lt is advisable to take these processes into account in order to examine their influence on the required storage volume. As the inflow data for this study had to be site-independent it has been decided t o transform the rain series into an inflow series in a rather neutral way, making use of our knowledge from the rainfall runoff process. Therefore a simple model, divided in a loss and a delay component, is applied (figure 27). ln,U Otipt.i Loss model Delay model Precipitation '------' Net precipitation '------' Inflow Figure 27 : The loss and delay component of the model In the following section first some restrictions and requirements for the construc tion of the model are given. Next a short description of each process and a sche matization for the model is discussed. Additional assumptions for the model parameters are based on literature data. A.2 Restrictions and Requirements For this model it is suggested to concentrate on paved surfaces in housing estates 43 The Des1gn of Infiltration and Percolation Facilities where, in many situations, disconnection is most obvious. In such areas roof surfaces, drives and other paved surfaces around the house are the dominant type of cover. The roof surface is assumed to be impervious so that losses caused by infiltration are zero. Interception losses on roof surfaces as well as runoff from or to unpaved surfaces are assumed negligible. The output of the model is an inflow series with the same format as the rain series. This makes a comparison between the design with and without losses and delay possible. Due to the delay process during runoff an overlap of two consecu tive rain events may occur. If this is the case, both separate rain events are com bined into one inflow event. The loss and delay process changes the starting time and duration of an inflow event in comparison to the original rain event. The model should be constructed in such a way that it adapts these variables. A.3 The loss process Precipitation losses depend strongly on the kind of pavement and the position of the area. For roof and paved surfaces the following losses are distinguished: 1. The initial loss 2. The loss by evaporation To remain on the safe side in the design process, the infiltration loss is neglected, as mentioned above. Initial loss The initial loss consists of two components: wetting loss and depression storage. The wetting loss is defined as the amount of precipitation absorbed by the pave ment. The absorption process starts at the beginning of a rain event. The wetting loss of roof surfaces has been researched by Pecher and Van de V en (9]. Their conclusions are given in table 2. Depression storage occurs if the wetting loss is exceeded by the supply of rainfall. On flat roofs this loss can be considerable, especially if runoff delaying facilities are applied. The slope of the surface also influences this loss. Van de V en ( 15] mentioned several values for the depression storage of roof surfaces. In the Wallingford procedure (91 a depression storage for inclined roofs is stated. Table 2 gives a summary of the literature data for depression storage. The third column shows the initial loss which is a summation of the wetting loss and the depression storage. 44 Appendix A. S1mulat1on of the prec1p1tat1on-runoff process Type of surface Wetting loss Depression storage lmtlal loss Source (mm) (mm) (mm) Flat surfaces: 0.2 0.5 0.2-0.7 0.4 - 1.2 Pecher (0 - 1 %) 0.5 2.0 2.5 vd Ven Inclined surfaces: 0.1 0.0 0.1 vd Ven (1 - 4 %) 0.4 0.4 Walhngford Table 2: Welting loss and depress1on storage (9]. Evaporation As long as a surface is wet, evaporation occurs. This can be considerable with a warm surface. The amount of evaporation from a paved surface is hard to quan tify. With the use of an energy balance, Pfeiff [9) calculated the evaporation of roofs due to the stored warmth in tiles. He found a value of 0.3 mm. Van de Ven [ 15] found evaporation values varying from 0.06 to 1.4 mm during a sunny su m mer day. He concluded that in most cases evaporation is lim1ted to about the open water evaporation. If the evaporation during the storm water event is ignored, the amount of ab sorbed water at the beginning of an event ( = wetting loss) will evaporate at the end of the same event. lt is supposed that all absorbed water evaporates before a new storm begins. Therefore the evaporation is assumed to be equal to the wet ting loss. Model assumptions The initial loss is considered as the only loss process, consisting of a wetting loss and a depression storage. The initial loss model is schematized as a reservoir in which the initial loss of the rain event is intercepted. The overflow from the reservoir represents the net precipitation. The reservoir is emptied after each inflow event by evaporation in order to intercept the initial loss of the next event (figure 28). F1gure 28 : Schemauzanon of the loss model 45 The Design of Infiltration and Percolation Facilities For the transformation of the rain series to inflow series an initial loss of 1 mm is used. Other losses that generally occur have been neglected, in order to be at the safe side. A.4 The delay process lt is clear that the precipitation is delayed during the runoff process. This delay is caused by two processes: • The initial loss of a rain event. This causes a shift in starting time of the event. • The runoff over the paved surface. This causes a flattening of the runoff wave. The modelling of the flattening of the runoff wave is described below. Model assumptions The simulation of the flattening of the runoff wave is based on a linear reservoir model. In this model a certain part of the net precipitation (pn (t)) is stored dynami cally on the runoff surface (h(t)) as the other part flows into the facility (q(t)): The inflow is linear proportional to the ratio of the dynamical storage and the reservoir coefficient: 1 q(t) =- ·h(t) (10) k in which: q (t) inflow [mm/si h(t) dynamically stored net precipitation (mm] k reservoir coefficient [s) Figure 29 shows a schematization of the delay model. 46 Appendix A. Simulation of the precipitation runoff process l p, (t) r------::.;------L .,., '------'------'~ F1gure 29 : Schemat1zat1on of the delay model A mass balance for this model yields: dh d t = Pn - q ( 11) in which: Pn(t) net precipitation [mm/si Together with equation 1 this yields: d h h ( t) d t = Pn - k (12) This equation integrated in time yields: h(t) = f The solution of this equation is: - !.. (14) k h ( t ) = h0 * e The emptying of a linear reservoir has a characteristic time, known as the reser voir coefficient k. After k seconds equation 14 shows: 1 = = h ( t k) h0 * e - 0.37 * h0 (15) 47 The Design of lnfiltrat1on and Percolation Facilities This means that after k seconds the dynamical storage is decreased with 63 % compared to h0 . In the next k seconds again 63 % of the remaining storage is discharged. For t = "" the reservoir is empty (figure 30). h (t) h , 0.37 h0 t -> ~ t =k t =2 k F•gure 30 : A characteristic emptying curve of a linear reservoir. For the model calculations the differential equation 3 is discretised as follows: (16) ~ h h => ~ t = Pn - k (17) => h(t +~t )-h(t) (18) => h ( t + ~ t ) = h ( t ) · ( 1 - ~ ) + Pn · ~ t (19) k Equation 19 is used for the simulation of the delay process. In this discretised formula the time step ~t has to be assessed, as the characteristic time k is a fixed parameter. The coefficient k depends on the length, roughness and slope of the surface. Van de V en (1989) recommends a value of approximately 5 minutes ( = 300 seconds) [15). As this value is calibrated for a housing estate, it is proposed to use it in the model. Consequently time step and characteristic reservoir coeffi cient are equal: after each time step 63 % of the dynamically stored water is discharged as the actual inflow into a facility. For equation 19 this yields: 48 Append1x A. Simulation of the precipitation-runoff process h ( t + /:::,. t ) 0.37 · h ( t ) + Pn ' /:::,. t (20) After the end of the rain event the reservoir will discharge the remaining water according to: h ( t + /:::,. t ) 0.37 . h ( t ) (21) As shown in figure 30 the water level in the reservoir will be zero at t = ""· An interruption criterium is used in order to determine at which moment the inflow event ends. This criterium assumes the reservoir to be empty if the water level is equal to or less than 1 ·1 0 -e m. A .5 The combined loss and delay model Summarizing, the complete loss and delay model is schematized as shown in the figure below. Model assumptions: Precipitation Evaporation - Initial loss ,.. 1 .0 mm - k ~ 300 s _j_----- _1_- -~ "" ''~''""''" Initial I '\. ,... r-,;, ------l l-.1..-----~ Inflow F1gure 3 1 : The complete loss and delay model The model is used to calculate the inflow series in the following order: 1 . The loss and delay reservoirs are empty. 2. The starting time, date and duration of a rain event is read from the rain series. 3. The loss model extracts the initial loss from the event. If the total precipita tion depth of a rain event is less than or equal to 1 mm, the event is ig nored and step 1 and 2 are repeated for the next rain event. 4. The new starting time and date of the inflow event are calculated. 5. The delay model transforms the (remaining) rain figures of the event into inflow figures until the reservoir is empty (the water level in the reservoir is 49 The Des1gn of Infiltration and Percolation Facilities less than 1· 10 .s m). 6. The ending time and date of the inflow event are calculated. 7. The computer program checks whether the inflow event, due to the delay process, overlaps the next rain event. If this is the case, this rain event will only run through the delay model as the initial loss of this event is ignored. This procedure is repeated until the rain series is completely transformed into an inflow series. A.6 Output The following graph shows the transformation of a rain event into an inflow event (figure 32). it is noted that the graph shows the course of the precipitation and inflow of one event. From this graph it is clear that during the first 150 minutes of the rain event a loss of 1 mm occurs: the inflow (dotted line) is equal to zero. The remaining part of the graph shows the transformed precipitation: due to the simulated delay process a peak reduction and a shifting in time of the peak itself is visible. In the next graph (figure 33) especially the loss process is visible. 50 Append1x A. S1mulat1on of the prec1p1tat1on-runoff process From precipitation to inflow 15 I I I ! I I I I I I I t I -11 I I I 1 ' ~ ~ i ~ 05 I 4: ' I ! i ' '. '- -1. 50 100 150 200 250 300 350 400 Tl"re (mn) lntl•l tos• 1 mm Ruerve~r c::oetl a ert • 300 sec F1gure 32 : Prec1p1tatton and Inflow vs. t1 me (23-02-19791. From prec1pttatac:n to n flow oa r------r------~r------~ I o• ! ~ 04 100 150 200 250 h•attows• t mm Re ...or c::oetlo«t • 300 EC F1gure 33: Prec1p1tatton and Inflow vs. t1me (02-01 -1979). I 5 1 The Des•gn of Infiltration and Percolation Facilities A. 7 Runoff coefficient The relation between the amount of precipitation and inflow is given by the runoff coefficient C [-). lt is defined as the ratio of the total inflow depth Pn [mm) and the total precipitation depth P [mm) on paved surface: C = pn (22) p For each rain event this coefficient is determined by the size of the event and the occurring losses. As stated before these losses depend strongly on surface charac teristics. Several literature sources mention runoff coefficients for roof surfaces. Both The Commission for Sewerage and Water Pollution and Kibler recommend values varying from 0.75 to 0.95 [10). Van de Ven assessed values from 0.45 to 0.85 [15). The consultancy bureau Tauw Civiel en Bouw uses a runoff coefficient of 0. 7 for the design of disconnection facilities in Enschede [5). In table 3 an average runoff coefficient is calculated for each year of the rain series and the additional calculated inflow. The second column shows the total precipitation depth per year, the third column the inflow depth and the fourth column the runoff coefficient. From this table appears that the yearly averages of the runoff coefficient vary from 0.64 to 0.80. The 15 year average is equal to 0. 74. These values agree with, or are slightly higher than the observed values by Van de Ven [15]. lt is therefore proposed to use the model with the assumption that for every rain event an initial loss of 1 mm occurs. If an inflow event, due to the delay process, over laps the next rain event, no losses appear. This loss model seems to be at the safe side for dimensioning the required volumes of infiltration and percolation facilities. 52 Appendix A. S1mulat1on of the prec1p1tat1on-runoff process Year Prec1p1tat1on Jmml Inflow lmml Runoff Coeffic1ent [-1 1970 749.4 543.3 0.72 1971 515.8 364.4 0.71 1972 707.8 492.5 0.70 1973 712.6 519.2 0.73 1974 706.4 512.7 0.73 1975 578.4 4 26.2 0.74 1976 406.2 261.8 0.64 1977 706.6 503.0 0. 71 1978 693.6 498.0 0.72 1979 924.9 733.1 0.80 1980 748.1 54 9.1 0.73 1981 932.0 710.3 0.76 1982 689.6 502.6 0.73 1983 912.3 714.5 0.78 1984 866.7 689.9 0.80 I Total: I 10850.7 8022.6 0.74 Table 3: Calculated yearly averages of the runoff coefficient 53 Appendox B. Extreme value analysis Appendix B. Extreme value analysis B. 1 Introduction Infiltration facilities are designed for rain events which are unlike to take place. One is, for instance, interested in the amount of rainfall {during e.g 15 minutes or 1 day) which is exceeded by an average of once per T year. The quantity T is called the return period. it is defined as the average length of the time interval between two equal events. Thus an event with a return period of 5 year occurs on average 20 times in a time span of 100 years. These return periods can be found by analysing the available series of data. The purpose of this analysis is to collect events with extreme values. In the traditional design procedure for disconnection facilities precipitation data are used. In our model the calculated water levels in the facility are subjected to extreme value analysis. 8.2 Theoretical background [3] In the hydrology two different statistical methods are commonly used for the analysis of extreme values. The first method is the extreme value series or annual maximum method. In this method only the largest value per year is collected. Thus a series of precipitation data of 50 year will give an extreme value series of 50 peak values. As only the largest value is collected, some information w ill be lost. If, for example, in one specific year 3 rain events occur with an extreme amount of total rainfall, only the highest peak value is taken from the series. The other two are not taken into consideration although it is possible they are larger than coll ected peak values from other years. This method is appropriate for the estima tion of events with a great return period { 1 0 - 1 0,000 years). The second method is the partial duration series. This series is formed by consid ering only the peak values from the original series, exceeding a certain threshold value. The value of the threshold is chosen in such a way that from the original series an average of 2 or 3 peaks per year remain. This method is better suited for return periods up to 25 years. As this is the range of return periods in which disconnection facilities are designed, the partial duration series has been chosen as the method to be applied. 8.3 The probability distribution of partial duration series The number of times that of the threshold value per year is exceeded can be described as a Poisson probability distribution. To which extent the peak values exceed the threshold can be seen as an exponential probability distribution. To- 55 The Desogn of lnfoltratoon and Percolatoon Facohtoes gether, the probability of appearance of a peak value with a certain size, meets the modified Gumbel probability distribution: Pr (x s ex(y-JJ)) = exp( -exp( - ex(y-JJ))) (23) In equation 23 a is a scale parameter and JJ is a place parameter. Estimating ex and JJ determmes the probability distribution. This is done with the following formulas: (24) - q ex N JJ 2. In 'A (where !!..) (25) ex r in which: N total number of peaks q threshold value amount of years from the data series y, = peak value The preferred value for 'A is about 2,0<'A<3,5 [15). Equations 23, 24 and 25 yield: Pr (x - q s ex(y - Jl)) (26) If a and JJ are known, it is possible to assess the y-point at the x-percent point of the distribution. For example, the 50%-point gives the y-value which is exceeded once in two years, the 90%-point gives the y-value which is exceeded once in ten years etc. Note that: T = ----- (27) 1 - Pr (x s y) 8.4 The use of the partial duration series in our model T he use of the partial duration series is allowed, provided the collected peak 56 Appendix B. Extreme value analysis values are independent events. For example, if the peak discharges of the river Rhine are analysed by means of the partial duration series, the values exceeding the threshold will be found to be in groups. This is caused by the strong depend ence between discharges of consecutive days. Only the highest discharge of this group of consecutive peak values should be considered for the partial duration series. If a disconnection facility is designed with the use of rain series (as proposed in chapter 2) , the same problem is encountered. Therefore a definition is required about w hich series of water levels can be defined as a single, independent event. From this event only the highest water level is added to the partial duration series. Definition: A single, independent event is a series of consecutive, interdependent water levels situated in the time interval between the moment the storage facility is empty and the moment the facility is empty again. lt is possible that two separate rain events result in one independent series of water levels. From this series only the highest value is added to the partial dura tion series. An example is given in figure 34 describing a fictitious event. In this figure only the peak water level B (25 ems) is added to the partial duration series. At the beginning of rain event 2 the facility is not empty. Thus the water levels resulting from two separate rain events can be seen as one independent event. Event 1 Event 2 e 0 ~ ! c c a: 2 E ...... :B. 0 30 ~ u ! 20 £ c > c -;: 10 ; 3: 0 0 2 4 6 8 10 12 14 16 18 20 Time (hours) Figure 3 4 : Def1nit1on of an Independent event 57 The Destgn of lnfiltratton and Percolatton Factltttes The input of our model consists of 15 years of inflow data. For a proper use of the partial duration series an average of about 3 peaks per year is needed. There fore the 50 highest peak values are collected, 46 of which are used. Normally the series of peak values is known. and after the threshold is fixed the number of peaks exceeding the threshold is counted. In our model a valid variant is used. The number of peaks is fixed (45) and the threshold is set as the average of 'peak 45' and 'peak 46' in the series of 50 collected peaks. If an mdependent event is found, the highest water level during that event is ranked in the output file with the 50 highest peak values. it is noted that the reservoir is considered empty if the water level in the facility is 1 cm or less. As the estimation of the parameters determines the probability distribution, for mula 23 may be rewritten. Instead of the calculation of a return period by a specific extreme value, the water level belonging to a specific return period is calculated: Pr = exp( - exp( - a(y - p )} Pr = e -e atr , , (28) ln[---1-1 - 1 In( Pr) (29) + J.1 y In( Pr) a The following formula is found giving the water level that will occur in the facility with a given probability for a certain return period: 1 ln[--- -1 In( Pr} (30) water/eve/ • y r = + J.1 + q a The probabilities of the required return period can be calculated with formula 27. Return period Probability 2 0.5 5 0.8 10 0.9 Table 4 : Return penods wtth the accessory probabtltttes For the construction of the design graphs the water levels, belonging to these three return penods mentioned in table 4, are calculated. 58 Appendix C. TOndern Appendix C. Tundern 111 Introduction In order to obtain a practical view of the application of infiltration and percolation facilities, a case study of the city of Hameln/Tundern (Germany) is presented below. In this city the quality problems in the receiving surface water caused by combined sewer overflows have been solved by the disconnection of impervious areas and the construction of infiltration facilities. The project executed in Tundern has proved to be a success according to everybody involved. Problem analysis Tundern is a small village near Hanover with approximately 2,300 inhabitants. The combined sewer system in Tundern was overloaded hydraulically. During heavy rainfall pavements and basements were flooded. Simultaneously an unacceptable high pollution load is released from the sewer overflow onto the river Weser. The connection of new impervious areas to the sewer system as well as the condensa tion of existing impervious areas have caused this overload. The total impervious area connected to the sewer system is about 29 hectares. This amount has in creased over the last decade from 25 to 36 % of the total area. In figure 35 the various parts of the total area in Tundern are shown. Total area 80 ha 7,496 • Pervious Areas (64.5 %) D Streets (12. 7 %) [l Roofs (15.4 %) 0 Yard (7.4 %) F1gure 35 : The total area in TOndern 59 The Design of Infiltration and Percolat1on Facilit1es Solution The construction of traditional in-line storage basins was the first option. This concept would cost about 5 million OM. The construction of the storage basins would destroy the good condition of most of the streets in Tundern. Because of its high costs and the difficult construction an alternative option has been search ed for. The spacious setup of the estates in Tundern made decentral infiltration facilities a possible solution. German regulations allow for a reduction of the "sewerage tax" in case a house-owner is able to drain his rainwater himself. The facilities had to be constructed on private lands and therefore voluntary coopera tion of the citizens of Tundern was needed. Preliminary research A precise survey of the area showed that the construction of decentral infiltration facilities was possible at almost 90% of the estates. The permeability of the soil (K-value = 1-5 1 o· 6 m/s) and the maximum ground water level (2.5 - 3 metre below surface) allow the use of this concept without restrictions. The efficiency of the proposed measures has been translated into the pollution load on the river Weser. The behaviour of the sewer system has been simulated with a computer program and 11 year of rainfall data. These simulations showed that the discon nection of 35% of the impervious area would give a satisfying result. The limit value of the pollution load onto surface water in the state of Niedersachsen is 250 kg COO/ArnP•"•'ous *year. To reach this limit the disconnection of 22,000 m 2 impervi ous area would be sufficient. Figure 35 shows the relation between the amount of disconnected area and the pollution load. The facilities The inf iltration of the runoff water will take place in grassed swales of 20 to 30 centimetres deep. The runoff from roofs and yards is led to the facility by hollow gutters constructed of natural stone. The down spout of the houses is cut off near the bottom and connected to the gutters. The facilities have mainly been constructed on private estates as 2/3 of the disconnectable impervious area is found here. The facilities have been built at some distance from the houses to protect the basements from ground water. The facilities have been designed in close deliberation with the estate owners. lt should be noted that for every estate a specific solution and design had to be found as the layout of the gardens and houses was in no two cases the same. To enlarge the acceptance an once-only investment subsidy of 10 OM per m 2 disconnected area has been offered to the estate owners who participated in the project . Moreover, their sewerage taxes have been lowered. In Germany these taxes consist of a domestical waste water 60 Appendix C. Tundern tax and a rain water tax. The rain water tax is calculated in ratio with the amount of impervious area of the estate. Pollution load {kg cod/A,.,,..,..Hw# • year} 290 -r------, 280 270 260 250 240 230 220 210 200 -t----r---,----,----,----.----~---r---- 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 Disconnected impervious area /ha] Figure 36 : Relation between disconnected area and pollution load In May 1995, after a period of 2 years of deliberation and construction, 160 estates have been disconnected from the sewer system. Some 1 00 more are in the design phase and will be constructed soon. The total costs of the project amount to 2 million DM. This is half the costs of the traditional storage basin solution. Some photos of facilities in Tundern are shown on the following pages. 61 The Design of lnftltration and PercolatiOn Factltties Ftgure 37 : A swale in a garden in TOndern for the infiltration of rainwater. In this estate both the driveway (front) and the roof have been disconnected from the sewer. The hollow gutter of natural stone is also clearly visible. Figure 38: This picture shows an tnlet pipe for rainwater into a facility. Some stones have been placed directly beneath the pipe to prevent erosion. In this case the facility has been constructed directly next to the building. This is possible as there is no basement under the building. 62 Appendix C. TOndern Figure 39: Here the creativity of the participants is demonstrated. it shows that the concept of making the rainwater visible at the surface (instead of discharging it in pipes below the surface) increases the involvement of civilians, in water management systems. Figure 40: Another example of the fine architectural skills of a participant. The water is stored in the water-butt in the upper-left corner. The overflow of this water-butt enters the small facility between the stones. The driveway has been disconnected by removing two of the small poles of the little fence between the garden and the driveway. 63 The Design of Infiltration and Percolation Facilities F1gure 41 Here the facility is still under construction. Note the very simple reconstruction of the down-pipe. it has been cut off at the bottom and a corner-piece has been con nected. This facility infiltrates only half the runoff water from the roof. As in most cases, the other half is discharged to a facility in the back-garden. 64 Append1x D. The design of the Ruwenbos M A-system Appendix D. The design of the Ruwenbos MR-system Introduction In the south-west part of the municipality of Enschede (The Netherlands) a new residential area, called Ruwenbos, is being developed. During the design process a lot of attention has been paid to durability. This has led to a remarkable approach of the water management system of the area. A storm drainage system was developed in which the roof areas and some of the roads are disconnected from the storm sewer system. The runoff from these areas is led to several infiltration facilities. The domestical wastewater is discharged by a separate sewer system to the wastewater treatment plant. As the infiltration facilities will only fill up during heavy rainfall, Enschede gave them the name 'Wadi-system', a reference to the subtropical rivers which only discharge water during and shortly after a rain event. The Wadi-system is derived from the German Mulden-Rigole system. In this appendix a summary of the design of the Wadi-elements is presented. This design was performed within the framework of a feasibility study for the applica tion of the Wadi-system in Ruwenbos [5]. lt should give an impression of the traditional design procedure used for the dimensioning of the Mulden-Rigole system. Finally a comparison is made with the design procedure presented in this report. Characteristics of the area The ground water levels are relatively high in winter and low in summer. The average high groundwater level (GHG) is less than 40 ems below surface, and the average low groundwater level (GLG) is more than 120 ems below surface. The GHG characterizes the December-February situation and the GLG the July-August situation. The area needs groundwater drainage in winter to create appropriate conditions for building and living. During summer the ground water leve ls allow infiltration of storm water. Nowadays the area is drained by the use of several ditches and canals which end in the Zweringbeek. The design capacity of this stream is 0. 731 m3 /s which equals 1. 7 1/s/ha. Traditional design of the elements The traditional design of the Mulden-Rigole elements for Ruwenbos is based on Intensity-Duration-Frequency curves. Assumptions for the design are given below: The rainfall runs off without delay and transformation. The runoff coefficient is 0, 7. 65 The Design of Infiltration and Percolation Facilities The infiltration capacity of the soil is a constant value in time. Discharge through the drains is neglected. The Mulden is modelled as a straight trench with infiltration through the bottom and not through the sides. This means that in reality the amount of infiltration will be larger. The Mulden is dimensioned in two steps, with for each step a different return period: -The Mulden without the slokops (d): return period of 2 years. -The Mulden including the slokops (d'): return period of 5 years. Figure 42 shows the different depths used in the design of the Mulden . ------...••• -·.-••it ··••-=- • •• F1gure 42 : Different depths used in the design of the Mulden The infiltration capacity of the bottom of the Mulden is set at 0.2 m/day. This is below the value found in the so il-research. The reason for this is that the infiltra tion capacity decreases during a rain event to this minimum value. If the soil is saturated, the water will flow down with a hydraulic gradient 1 m/m. If the soil is unsaturated the hydraulic gradient can be significantly larger due to the suction stresses of the soil matrix. By taking the infiltration capacity at 0.2 m/day, the discharge can be taken at the same value as the infiltration capacity. If the water level in the Mulden reaches level d, the slokops starts to discharge water to the Rigole. The infiltration capacity of the slokops is 100 m/day. This is a relatively low value for gravel but the slokop could get clogged partially. Standard Terrain for the design From the layout of Ruwenbos a standard terrain was chosen for the design of the elements. This terrain has a relative large impervious area. Therefore the element designed for this area can also be applied in other areas with a smaller impervious area. The total connected impervious area to the standard element is 1,945 m 2• The length of the element is set by the layout of the area at 51 metres, the width of the Mulden at 3 meters. This means that the total surface of the element is 7 to 8 % of the connected impervious area. 66 Appendix D. The design of the Ruwenbos MR-system The Intensity-Duration-Frequency curve according to Buishand and Velds is used for the design. The y-axis of the curve is transformed from the rainfall intensity into the amount of water discharged to the element with the following formula: V,o, = P * 1 0·3 (A ompv * a + Amuldenl In which: amount of water discharged to the element [m3] rainfall intensity [mm] connected impervious area 1m2] surface of the Mulden [m 2] a runoff coefficient [-] If the width of the Mulden is chosen, it is possible to find the construction depth d from the adapted Intensity-Duration-Frequency curve. The amount of water stored in the Mulden (Amulden * d) together with the amount of water infiltrated (~ ulden * infiltration capacity * t) must be equal to the amount of water discharged to the element. The determination of d can be seen in Figure 43 below. 3 Amount of water (m ) Time /hi F•gure 43 : Determination of d From the adapted Intensity-Duration-Frequency curve with a return period of 2 years can be read that the depth of the Mulden should be at least 0 .23 centi metres. The emptying time is t hen 30 hours. If the water level in the Mulden is < level d, the following equation is valid: 11 storage = inflow - k mulden * Amulden * t i If the slokops start discharging water (t1 ) the next equation becomes valid: 11 storage = inflow - (kmulden * Amuldon * t) - (kslokop * Aslokop * (t-t, )) 67 The Des1gn of lnf1ltrat1on and Percolation Fac1ht1es in which: A.nvlden total surface of the Mulden [m2] A slokop total surface of the slokops [m2] kmulden the infiltration capacity of the Mulden [m/s] k slokop = the infiltration capacity of the slokop [m/s] At both ends of the element a slokop is placed with a diameter of 0 .5 metre. The 2 total area becomes then 0.4 m • If the total outflow from the second formula is plotted in the adapted Intensity-Duration-Frequency curve with a return period of 5 years, the total depth of the Mulden can be determined (see figure 44). Return period - 5 years 66.03 l so 52.82 : ..s .i 60 39.62 ! "( " E :::: .:::.. .:! .. 40 2 6.41 .s:.. ...:; 0:: -'!" 20 13.21 "' 0 0 0 120 240 360 480 600 720 840 960 1080 Duration (min) 1L. _ _ H_-_o_. 1_m_ __H_-_ o_.o_1_4_m-'l F1 gure 4 4 : Det erm1 nauon of Mulden depth for a return penod of 5 years lt shows that the necessary construction depth above the slokops is 0 .014 metre. This construction depth is negligible. If h is set at 0.1 metre, the rain event with a return period of 5 years will not present any problems. lt can be said that the Mulden is overdimensioned. The rain event with a return period of 25 years w ill lead to an overflow of the element into the next element. This is shown in figure 45. 68 Appendix D. The design of the Ruwenbos MR-system Return period - 25 years 10 66.03 ;;- ~ l 80 52.82 ! e • 60 39.62 ,~ " .:! ::.• ·! . 40 26.41 ~ "' ~ .!! " 20 13.21 0 0 0 120 240 360 480 600 720 840 960 1080 Duration (min) F1gure 45 : Control calculation with a return period of 25 years Within the design of the element, the drain discharge is neglected. The storage in the Rigole and the percolation from the Rigole into the surrounding soil are also neglected, leading to a very safe design. A Mulden with a width of 3 metres and a depth of 0.35 metre meets the required storage volume for the chosen standard terrain and consequently can be applied in the other areas as well. If the infiltration capacity decreases dramatically, an overflow will occur once in 5 years. Clogging of the slokop has no influence. The slokops are only used if the soil is highly impermeable. The layout of the elements can be chosen freely. Only the infiltration surface and the storage volume of the Mulden are important. The design graphs for the Mulden-Rigole system as presented in chapter 6 have the same starting points as the design of the Ruwenbos Wadi-system. If the design graph presented in figure 25 is used for the standard terrain of 1945 m 2 and a return period of 5 years is chosen, a construction length of 42 metres is found. The Wadi-length in the Ruwenbos design is 51 metres. lt was already known that this Wadi was overdimensioned and that the length of the Wadi is determined by the layout of the residential area. lt should therefore be noted that the comparison between the results of the two procedures only has an informative character. 69 COMMUNICATION OF TilE DEPARTMENT OF SANITARY ENGINEERING AND WATER MANAGEMENT In the series "Communication of the Department of Sanitary Engineering and Water Management" are edited: l. Siebers, II.H.: Pauems and variability of phosphate and heavy metals in sedirnents of two shallow lakes. (November 1985) 2. Flipse, M.J. en Heide, J, van der: Ontwikkelingen met betrekking tot vaste a!Valstoffen ex art. 4, 17, 25, 26 van de AIValstoffenwet in periode van ea. 1980 tot 1985. 3. Kop, J.H.: Pla.nvorming voor de drinkwatervoorziening. (februari 1986) 4. Blankeo, J.G. den en Hoogh, M.P.A.J. de: ModeUen voor desinfeetie van gczuiverd a!Valwater met chloor en ozon. 5. Kop, J,H.: Het problcem van de wederzijdse afstcmming van de bclangen van dnnlcwatervoorziening en milieubcscherming bij de planning voor de winning van zoct grondwater. (augustus 1986) 6. Boekelmao, R.ll. eo Niet, H. de: Het bcrckenen van modelkrommen voor Geo-eleklrische mctingen. 7. Vos, W.L., Donze, M. and Buiteveld, H.: On the reflectance spectrum of algae in water: the nature of the peak at 700 nm and its shlf\ with varying algal concentration. (October 1986) 8. Smit, D., Mllmeren, H.J. van en Veldkamp, R.G.: De zuurstofbuishouding van de Utrechtse Vecht. (november 1986) 9. Heide, J, van der: Kinctische modcUen voor ontwerp en bchecr van actief-slib-installaties deel I en 2. (februari 1987) 10. Boulao, R.P., Donze, M. eo Klllpwijk Sj.P.: Fosfaatbalans van de polder Recuwijk en ccn aantal deelgebiedcn. (maart 1987) 11. Groot, C.P.M. de eo Breemen, A. N. van: OntspanningsOotatie en de berciding van drinkwater. (juli 1987) I2. Blllokeo, J.G. den eo IIoogb, M.P.A.J. de: Modelvorming voor verwijdering van indieatororganismen in het aetief-slibproces. 13. M.isbra, K.K. and Breemen, A.N. van: Gravel-bed Oocculation. 14. VIis, E. van der: De ftltratiethcorie. (maart 1988) 15. Koremao, E.A. en Breemeo, A.N. van: Toepassing van het vriesdooiproces bij de ontwatering van coagulatieslib. (mei 1988) 16. Gaozevles, P.P.G., Kop, J.ll. en Ywema, R.: Materiaalkeuze a!Valwaterleidingen. (juni I 988) 17. Nieuwenbuyze, R.F. van, St.okmao, G.N.M., Kuijper, R., Gerritseo, J,J. en Donze, M.: Detectie van proceswater met behulp van thermische remote-sensing. (juni 1988) 18. Blaokeo, J.G. den eo IIoogb, M.P.A.J. de: Modelvorming voor een goede procesregeling van de desinfectie met chloor c.q. ozon aan de hand van instelbare en/of direct meetbare variabclen. (augustus 1988) 19. Noppeoey, R.M.: De invloed van stagnante zones op dispersie. (november 1988) 20. Noppeoey, R.M.: Gevoeligheidsonderzoek Alarmmodel Rijn; De invloedslcngte van samenvloeiingen bij dispersie. (november 1988) 21. Noppeoey, R.M.: De verspreiding van olie op rivieren benadcrd met het Taylor-model. (novembcr 1988) 22. Noppeoey, R.M.: De invloed van near-field processen op ecn far-field dispersie beschrijving. (november 1988) 23. EUen, T. van : De invloed van afvoerfluctuaties op de verspreiding van een verontreinigingsgolf. (juni 1989) 24. Blanken, J.G. den: Afscheidssymposium prof.ir. A.C.J. Koot. (januari 1989) 25. Hooykaas, L.J., Donze, M. en Klapwijk, Sj.P. : Fosfaatbalans van de polder Reeuwijk en de Rceuwijkse plassen. (januari 1989) 26. Verwoerdt, P. en Mazijk, A. van; De ccn-dimensionale dispersievergelijking van Taylor bij een opdeling van de rivier in vakken. (maart 1989) 27. Mazijk, A. van: Gevoeligheidsonderzoek Alarmmodel Rijn; eindrapportage. (mei 1989) 28. Blanken, J.G. den en Hoogb, M.P.A.J . de: Desinfcctie van behandeld afvalwater met chloor: vergelijking van eenpunts- en tweepuntsdoscring; deell: Tekst, bijlage A, B en C. deel 2: Bijlagc D, E, Fen G. (mci 1989) 29A. Verstappen, G.G.C.: Gedrag van organische micro-verontreinigingen in rivieren. (juli 1989) 298. Mooren, j.J.M. en Heide, J. van der: Leaching of heavy metals from lhermaUy decontaminated soils. (maart 1989) 30. Nieuwstad, Tb.J., Wortel, N.C., Bout, F.N. vao deo eo Alting, B.J.: Een vergelijking tussen ladingsgewijze en continue zuivering van afvalwater. (juni 1989) 31. Kramer, J.P., Wouters, J.W. en Kop, J.H.: Dynasand Filtratie. (July 1989) 32. Nieuwstad, Th.J.; Treatment of municipal wastewater in a pilot-scale airlift-loop reactor. (December 1989) 33. Ankum, P.; Polders; achtergronden, ontwerp en toekomstige ontwikkelingen. (juni 1990) 34. Brandsma, T.; Evaporation in Hydrology and Meteorology. {July 1990) 35. Mooren, J.J.M.: Het uillooggedrag van kunstmatig samengestelde en verontreinigde grond. (2 delen) (augustus 1990) 36. Siogb, S.N., Boekelman, R.H., Rientjes, T.H.M. eo Dam, J .C. van; Behaviour of ground water of the polder Groot-Mijdrecht. 37. Boekelman, R.ll. eo Rientjes, T.H.M.; Workshop hydrological models. (September 1990) 38. Stavrides, N., Rientjes, T.H.M. en Dam, J.C. van; Network optimization, a simple approach applying GIS and MLR. 39. Duindam, P., Morales, C. and Heide, J. van der; lnvestigacion sobre Ios desechos solidos de la ciudad de Masaya, Nicaragua. (enero 1991) 40. Heide, J. van der: Evaluacion hidrauliea de plantas potabilizadoras de ftltracion rapida en Nicaragua. (junio 1991) 41. Heide, J. van der: Metodologia de potabilizacion de agua superficial en Nicaragua. (abril 1992) 42. Veldkamp, R.G.; Randvoorzieningen van rioolstelsels kritisch beschouwd. (oktober 1991) 43. Kruithof, J.C., Scbippers, J.C. and Dijk, J.C. vao: Abastecimiento de Agua Potable de Agua Superficial en Ios aiios Noventa. (febrero 1992) 44. Rietveld, L.C. and Mal~inhe, N.P.: Pilot Plant Studies on Slow-Sand-Filtration and Up-Flow-Roughing Filtration in Mozambique, (January 1993) 45. Brandsma, T.: Evaporation and Climate Change. (February 1993) 46. Oude Essink, G.H.P.: Sea Level Rise: What are the causes of temperature changes and how is sea level rise related to temperature rise bol11 in the past and in l11e future?; a literature survey. (July 1992) 47. Xiaodi Hao and Nieuwstad, Tb.J.: Feasibility of simultaneous nitrification and denitrification in a pilot-scale airlift-loop reactor. (June 1993) 48. Brandsma, T.: Sewer Systems and Climate Change. (September 1993) 49. Ankum, P. and Brouwer, R.: Nevengeulen en Sedimentverdeling. (November 1993) 50. Brouwer, R.: Review of the water management systems in the Gujarat Medium Irrigation 11 Project. (August 1993) 51. Hoornstra, J.S. and Jong, J. de: Water quality management in the Netherlands. (October 1993) 52. Jong, J. de: Third policy document on water management (June 1993) 53. Jong, J. de and Hoornstra, J.S.: Analyse der anforderungen im EG-bereich (October 1993) 54. Jong, J. de, Baarsma, J.P., Frintrop, P.C.M. and Mulder, W.H.: Screening strategies for organic micropollutants in context with water pollution control. (October 1993) 55. Jong, J. de: The method and mechanisms of establishing consensus on water management policy in the Netherlands. (October 1993) 56. Oijk, M.J. van and Rientjes, T.H.M.: Geostatistics and Hydrology; part I: Spatial and Temporal Variability. (September 1993) 57. Oijk, M.J. van and Rientjes, T.H.M.: Geostatistics and Hydrology; part 2: Estimation Techniques. (November 1993) 58. Oijk, M.J. van and Rientjes, T.H.M.: Raingauge network optimization and GIS; a case study of the Mananga Basin. (October 1993) 59. Oijk, M.J. van and Rientjes, T.H.M.: Geostatistics and Hydrology; part 3: Hydro-Meteorological Network Design. (May 1994) 60. Nieuwstad, Tb.J.: Modellering van cen internal-loop airlift reactor Watersnelheid en gas- en dceltjesfracties. (Juli 1994) 61. Koning, J. de: Cyaniden - eigenschappen, analyse en behandeling van cyanidehoudend afvalwater. (olctober 1994) 62. Koning, j. de: Mcetrapport bij het onderzoek naar de gekombincerde behandeling van met hexacyanoferraat verontreinigd grondwater en stedelijk afvalwatcr. (olctobcr 1994) 63. Buureo, J.C.L. van and lleide, J. van der: Abatement of the Water Pollution in the lnfulene Basin; Final Report. (February 1995) 64. Blankeo, J.G. den: Modellering (effiuent)fLitratie. (augustus 1995) 65. Ankum, P.: Flow Control in Irrigation and Drainage. (June 1995)