Flushing with gate valves

corrosion and pollution control was published by Fan et al. (2001). A case study of the sewer and drainage flushing system in Cambridge, Massachusetts shows the successful application the HydroSelf flushing system. [Pisano et al., 2003] Later results of a case study in Cambridge, Massachusetts and a cost analysis are presented by Fan (2004).

Flushing with gate valves 107 To optimise the flush cleaning with a gate valve the following list of recommendations are formulated by Drews:

• To create an optimal front wave the width of the gate opening a in ratio to the pipe diameterD should be chosen according to the following values:

0,23≤ a

D ≤0,26 (7.1)

• The cleaning effect is favoured by a fast opening of the gate valve.

• The cleaning effectiveness could not be improved when the storage length exceeded 300-fold the diameter of the sewer pipe. (forI_s = 1 : 1500 to 3.000)

• The cleaning effectiveness is constant and independent of the storage length for subcritical flows.

• A complete opening of the gate valve increases the cleaning effectiveness for sub-critical flows.

• The cleaning effectiveness of gate valves is comparative small for a low sewer slope and deposits with a high mineral content. Therefore the usage of gate valves is recommended only for carefully chosen locations in the sewer system with ideal boundary conditions.

Lorenzen investigated the effects of two gate valves in main sewer channels in the city of Hannover, Germany to prove their cleaning abilities in chosen sewer reaches. [Lorenzen et al., 1997] The sewer channel for mixed sewage had a length of 2400 m with a diameter of 1800 - 2400 mm. After 1340 m length it had a lateral inflow. One part of the sewer with a length of 50 m had a width of 4 m, a height of 2 m and was called Wide Chamber. The bottom slope of the channel was between 1.2 and zero, which led, together with four bends, to sediment accumulation after 1200 m length. The flushing device consisted of two, not described, gate valves. One was placed after 611 m and the second was installed in the inflowing channel. The storage height was 1.9 m with a storage time of 12 hours.

There was no information given concerning the storage volume. The opening time for the gate valves was given with 5 min. Due to the slow opening time the created flush wave can not be considered as a dam-break wave, which makes the results difficult to compare to standard flushing devices. The measured sediments showed a height of up to 20 cm in the wide chamber. The texture of the deposits consisted of sand at the bottom and muddy organic material in the top layers. The mean particle size was measured with d₅₀ = 0.6 mm. Additionally the water level and vertical velocity distribution were measured with an induction device. After the measurement of the initial conditions five flush waves were created before the sediment heights were taken for the second time.

The results showed, that for 600 - 700 m the flushing was effective. The deposits at the bottom of the channel were removed and transported with the wave. In the wide chamber only the organic material was moved by the flush wave. The heavier parts, sand and gravel, remained at the bottom. Even after further 24 flushes the situation did not change. The second part of the investigation was regarding the effect of the flush waves on the long-term cleaning of an initially high-pressured cleaned sewer. It became obvious that the flush waves created by the gate valves were not strong enough to keep the sewer free of sediments. Even the combination of both gate valves were not powerful

enough to create the necessary bottom shear stress to avoid sedimentation.

Lorenzen made several conclusions on the results of the flushing test. He stated that the cleaning success of a flush wave is depending on the storage height, the flow velocity and the duration of the flushing. He did not regard the storage volume but he gave the general statement that a long flush wave is equally effective as a short flush wave.

The turbulence and the shear stresses in the front wave are responsible for the lifting of the deposits. In contrast to his investigations Lorenzen stated that is more effective to create several consecutive flushes than single large flush waves. He also concluded that a remaining water body would degrade the energy of a flush wave and therefore the cleaning success. Flush waves will remove most of the biogenetic material of the deposits and this leads to less pollution of receiving waters during rain events. These investigations and the results were also published in Lorenzen et al. (1996).

Ristenpart (1998) carried out flushing test with gate valves in two parallel sewer channels (diameter 1500 mm) in Hildesheim, Germany. By measuring the water level, the flow velocity, the deposit level and the deposit volume before and after the flushing, he found that frequent but less intensive flushing is more effective than a single flush wave of high intensity.

Ristenpart was not able to clean the sewer on the whole investigated stretch using the flushing device. Only 6.30 m were totally cleaned. (Figure 7.5) He determined that for a certain deposit volume no further cleaning was possible, which would be due to a balanced condition of the deposits. The complete cleaning of a sewer with deposit bed would not be possible under the circumstances in Hildesheim, but keeping a sewer clean for a long duration by regular daily flushing after an initial cleansing could be achieved.

Figure 7.5: Change of the deposit level caused by flush waves [Ristenpart, 1998]

These results were confirmed by Gendreau et al. (1993), who investigated the effect of flush waves on sand in laboratory tests. Therefore Gendreau use a servo-controlled weir to create flush waves from a tank into a 24 m long flume to observe the motion of the sand. He determined that the sand is transported for a longer time when the weir was opened fast with a high storage level. The complete drainage of the storage tank (V = 4.8 m³) led to the best transport results for the sand particles.

Experimental and numerical investigations on the scouring effects of flushing waves on sewer sediment deposits were presented by Campisano et al. (2004). An experimental series was carried out in a laboratory channel using a simple flushing device and

consid-Flushing with gate valves 109 ering different boundary conditions. The main idea was to develop a simple model to describe the scouring effects of flush waves on a non-cohesive sediment deposited on the bottom of a sewer channel. The results of the numerical model were then compared with the results of the laboratory experiments to derive indications on the scouring processes due to the flushing operations.

The flush tests were carried out in a 3.9 m long and 0.15 m wide rectangular channel with a slope of 0.145 %. The flushing device consisted of a stainless steel plate, which could be opened very fast to create a dam-break flush wave. Volcanic sand was used as a substitute for sewer deposits. The one-dimensional numerical model was based on the Saint-Venant equations for unsteady flow and Exner equation for the sediment continu-ity. The bed-load transport of the sediments due to the unsteady flow conditions caused by the flush waves was calculated by the equations of Meyer-Peter and Kalinske. The Saint-Venant equations were solved numerically using the second order MacCormack scheme. This scheme is shock-capturing, which means it is able to describe discontinu-ities such as hydraulic jumps and shock waves over an initially dry bed. It is well suited to analyse the movement of flush waves. To reduce the numerical oscillations caused by high gradients in water level and flow rates the total variation diminishing (TVD) scheme was also applied.

The results of the numerical model for the progression of the flush wave were compared with the experimental data and showed a very good agreement up- and downstream of the flushing gate. The movement of the sediment was well reproduced for the first flushes but with an increasing number of flush waves the differences between measured and simulated results became obvious. Tests with changing boundary conditions and different variable for the Meyer-Peter and Mueller formulae led to better results. A good agreement was also found for the comparison of the washed out sediment weights and the scoured section lengths after a number of n successive flushes. To summarize the results it can be stated that the numerical model offered a good reproduction of the experimental data from the laboratory channel.

Campisano et al. (2005) also presented the results of a dimensionless numerical inves-tigation on the scouring performances of flushing waves. The invesinves-tigation has been performed using the numerical model mentioned before. It was validated on the basis of laboratory data. The evaluation of the dimensionless sediment transport rate Q_s has been derived from the Meyer-Peter and Muller’s formula. The flush waves were created by a vertical lift gate in a laboratory channel.

The experimental evaluation of the performances of the flushing waves was carried out by measuring the scoured flume lengths L_s, the distance between the initial position of the upstream deposit front and its position after n flushes. The bed profile evolution was also monitored by measuring the sediment heights h_s along the flume at the end of each flush. The numerical simulations have allowed evaluating the scoured channel lengths downstream of the flushing device, as a function of the number of performed flushes and for different values of the identified dimensionless parameters. The results have pointed out the increase of the scouring effects as the channel slope increases and the sediment grain size decreases. The results have also provided information on the design and set-up of the flushing devices, with a specific reference to the evaluation of the most effective hydraulic conditions of the flush. In particular, the results have shown a slight convenience in operating a small number of flushes with high water heads instead of frequent flushes with lower water heads. [Campisano et al., 2005]

The most recent investigation of Campisano et al. (2006) was carried out on flushing experiments with cohesive sediments. The aim was to evaluate the effects of fine cohesive sediments include into a granular sediment bed. Therefore a bed of sandy material was formed in a 17 m long laboratory flume as a reference condition for the first experiments.

Then a clay-silt mixture was added to the sand in measured values of 10 and 20 % weight content. The addition of the fine particles allowed keeping the initial experimental conditions equal to the one performed with purely sand deposits. The fine material was able to fill a part of the sand porosity without changing the initial height of the sediment bed. The critical shear stressτ_crit for the erosion of the sediment mixture was evaluated between 1.4 - 3.3 N/m² depending on the percental content of fine material. A vertical lifting gate at the upstream end of the flume generated the flush wave. The deposit bed evolution was measured as well as the sediment masses flushed out by the waves.

The observation of the erosion processes during the experiments showed the prevalence of bed-load transport for sand particles while the silt-clay material was observed to move mainly as suspended-load. It also became obvious that for the first flushes the addition of fine material slowed down the advancement of the sediment bed. The more the content of fine material increased the more the movement of the sediment bed was slowed down.

Differently from what was expected, with an increasing number of flushes, higher erosive effects occurred in beds with fine material. The removed sand masses were larger for deposits with fine material.

The experimental data derived from this investigation will be taken by Campisano for the validation of numerical models developed for the effects of flush waves on sewer deposits with cohesive properties. [Campisano et al., 2006]