SIPHON ANTI-CRUE

Abstract

To compensate for the rise in water level caused by an increased flood flow, the marine propulsion system of the turbine (6), nozzle (7), and engine (5) of the present invention draws water from a protected reach to increase its flow, lower its level (1-A'), and discharge it through a siphon into a downstream reach whose level (2-A') corresponds to the new imposed flow. The excess flow allows for the control of the flushing of upstream reaches and retention basins. In the case of the 1910 Paris flood (8.62 m), and with the help of these retention basins, a pumping power of 38 MW at a total cost of less than €1.7 million would have made it possible to maintain the water level at 5 m. 6 MW can keep the riverside roads dry during ordinary winters.

Description

-1- After periods of heavy rain or abundant snowmelt, three factors explain the floods that inundate cities: an excessive supply flow, resulting in a rise in the water level of the affected canal section, and insufficient downstream drainage capacity. Retention basins mitigate excessive flows, and dikes contain the water; the present invention extends their function by lowering the water level and improving flow. Climate change and the increasing frequency of these events demonstrate that the threat is not a thing of the past, and that no truly satisfactory solution has yet been identified. With land fragmentation, new agricultural practices, the proliferation of engineering structures on rivers, and thus the increased vulnerability of cities, the damage suffered is worsening. To describe the earlier technique, let's give credit where credit is due: Chinese mythology speaks of floods comparable to that of Noah's Ark; to remedy this, Emperor Yao asked Gun to stem the tide and regulate the waterways. The dikes and dams eventually broke, and Gun was put to death. His son Yu, known today as "Yu the Great," successfully took up the task by doing the opposite: deepening the beds of rivers and lakes, digging diversion canals, and cleaving through mountains to efficiently drain the rivers to the sea. In the West, this scourge is taken into account as Gun did by developing dams and retention areas upstream to reduce the flow by overstocking the water and delaying its flow, as well as by building permanent or temporary lateral dikes to protect areas to be protected from flooding. During the collection of solutions following the 1910 flood in Paris, Mr. Soulacroup, a schoolteacher in the Lot region, proposed hastening the flow of floodwaters using waterwheels submerged in the current. The need to dredge riverbeds to increase their capacity to handle high flows at a constant level was also mentioned. The answer to parliamentary question no. 47146 published on April 1, 2014 on page 3025 of the Official Journal, the book "The day the water returns" by Pascal Popelin, President from 2001 to 2013 of the board of directors of the great lakes of the Seine and the documents written following the floods of 2016 and 2018 indicate that the embankment of the banks and the extension of the peaking capacity by 55 million additional cubic meters constitute the only solutions considered to date, but prove powerless in the face of the flood of 1910.Mr. Popelin's international research, for example in the United States and the Czech Republic, led him to resign himself to containment and peak reduction. To explain the present invention, the "anti-erue siphon" starts from the observation that the measures traditionally put in place reduce the Froude number, a dimensionless number which characterizes a flow regime; this invention seeks on the contrary to increase it. As a reminder, a low Froude number (Fr=v / Jah, where v is the fluid velocity, g is the acceleration due to gravity, and h is a characteristic length) corresponds to a large water depth and low velocity (equivalent to subsonic flow) in a downstream-driven river regime: moving particles are constrained by those preceding them. This phenomenon is related to dynamic traffic jams on highways, which transform a slowdown into a complete standstill that propagates slowly upstream as it dissipates downstream. Conversely, a high Froude number corresponds to a torrential regime, with a small water depth and high velocity (equivalent to supersonic flow). In this regime, the fluid is pulled by the forces of gravity, without the downstream fluid mass causing any obstruction. The flood control siphon utilizes a second observation: that flood flows can reach the order of magnitude of those passing through the propulsion systems of large modern ships. The tens of megawatts required are within the range of available large electric motors and current electrical grids (powering high-speed trains). A third observation stems from the differentiation of the canal reaches, which are not all alike: some are highly vulnerable to flooding, others considerably less so. Those that are urbanized can hardly be modified, while others can accommodate significant flow rates, particularly by increasing their depth, also known as "drafting depth," if they are not constrained by numerous bridges and tunnels. Based on these three observations, the present invention consists of pumping water from an upstream reach and discharging it into the downstream reach, with the aim of lowering the level of the upstream reach, increasing the flow rate, and clearing the reaches further upstream. The traditional dam between the two reaches is used in reverse, since the imposed flow rate raises the downstream level above the upstream level.Regardless of these two levels, the use of inverted U-shaped pipes called "siphons" limits pumping power and therefore energy expenditure to the bare minimum. Pumping periods are coordinated with operations related to retention basins to judiciously distribute their impact over time. This involves, on the one hand, anticipating large flows by opportunely flushing the upstream network, and on the other hand, distributing them between upstream reservoirs, downstream reservoirs, and the sea's absorption capacity (tides, coefficients) to manage flooding across the entire catchment area and remain able to cope with other rainfall events. This invention leads to the paradoxical use of dams already in place to maintain a minimum water level during periods of low water. As illustrated in Figure 1, during floods the level (1-A) of the upstream reach (1) is lowered by means of one (or more) hydraulic pump(s) with a flow rate comparable to that of the reach. composed, for example, of a propeller (6) inside an excavation in the dam's foundation constituting a nozzle (7), which discharges into the downstream reach (2) through an inverted U-shaped conduit called a siphon (4). With the motor (5), the pump thus formed locally reduces the level downstream of the upstream reach, increases its slope (1-A°), its flow rate, its Froude number, reduces the average level, and can also flush out reaches even further upstream. Under the effect of the additional flow rate, the level of the downstream reach (2) will rise to (2-A°), where the new level and the new slope correspond to the flow rate imposed by the flood control siphon. This is a stable but paradoxical state during which the level of the downstream reach (2) is higher than that of the upstream reach (1) and forces the dam (3) to perform a mechanical response opposite to the periods when the flood control siphon is off, thus providing effective hydraulic retention in the opposite direction.Under normal circumstances, the waters of the upstream reach (IA) are prevented from flowing towards the downstream reach (2-A) and, when the hydraulic device is in operation during flood periods to raise the waters of the upstream reach (1-A°) up to the downstream reach (2-A"), the waters of the downstream reach (2-A") are retained to prevent them from flowing towards the upstream reach (1-A°), which is significantly lower. In the upstream reach (1), the water level will drop to level (1-A°) due to the suction of the flood control siphon, then gradually to level (1-A””) due to the additional flow. To avoid restricting the flow, the upstream dam (9) will be lowered: the flow in the upstream reaches will increase until it reaches that of the downstream flow. Their levels (1-A°*) will be subject to two opposing forces: dropping, like the level of the reach into which the flood control siphon draws water, and rising due to the additional flow. These levels (1-A””) will in any case be below level (1-A) in the absence of the flood control siphon. While the safety of people below level (1-A) may seem precarious at first glance, keeping property above water is already an undeniable achievement. Since the water levels in the two reaches can vary considerably depending on the flow rates, the connecting conduit fed by the pump takes the form of one (or more) inverted U-shaped siphons (4), one outlet of which, in the upstream reach, is fed by the nozzle (7). The other outlet of the siphon feeds the downstream reach. Spanning the dam (3), this siphon (4) recovers the energy corresponding to the drop between, on the one hand, the siphon's throat near the top of the dam and, on the other hand, the level (2-A°) of the downstream reach (2). This recovery is possible for all the respective heights of the upstream and downstream reaches, within the classic limit of approximately ten meters of atmospheric pressure reduced by head losses. It should be noted that the shaft between the motor (5) and the propeller (6) does not pass through the siphon (4) to avoid the use of a stuffing box, which would be necessary due to the low pressure at the top of the siphon.Therefore, there is another "U" between the nozzle (7) and the siphon (4) to redirect the flow upwards, with the largest possible radius of curvature to limit pressure losses. The capacity of the upstream and downstream reaches to accept the new flow imposed upon them without causing flooding is crucial; they must have been modified to reduce the rise in water level that this new flow will cause, for example by carrying out dredging operations. This new flow must be compared to a maximum flow, and therefore to a maximum water level, which must not be exceeded. If the flood control siphon is to be integrated into the upstream and downstream environment, it must also take into account the flood history. It turns out that floods have always been predictable with considerable advance notice, with only a few false alarms. Therefore, there are conditions under which the upstream retention basins deserve to be purged by the action of the anti-flood siphon without exceeding the flow rate maximum. When the flood arrives, the received flow rate can be increased, anticipating the quantities meteorologically and hydrologically expected. When the flood exceeds the maximum flow rate, we will see in the presentation of a detailed method realization that the flood control siphon can lower the average level of the millrace main flow without increasing the flow rate. The downstream flow rate is therefore not increased and the other reaches of the river do not see their situation worsen. The situation finally deserves to be managed in its entirety, geographically speaking, according to retention reservoirs upstream and downstream to the sea; the tides and their Coefficients determine the maximum acceptable flow rate near the sea. The paradoxical difference in water levels between the reaches will be maintained as long as the pump is operating. It is essential to prevent a pump malfunction from causing the siphon to operate in reverse, that is, emptying the downstream reach into the upstream reach and accelerating the rise of water in this protected reach. To this end, a siphon-breaking vent (8) described in Figure 2 would connect the upper part of the siphon to atmospheric pressure and instantly interrupt this untimely emptying. The upper part of the inverted U-shaped conduit (4) of the flood trap houses the siphon break vent (8) described in Figure 2. The vent (12) is a tubular branch (11) at the top of the trap (4), terminating in a horizontal section (13). A stopper (14) is fitted to this section, along with the necessary sealing gasket. This stopper is mechanically connected to a lever arm (17) equipped with a counterweight (18), and to a paddle (15) that intercepts the flow within the trap. The paddle (15) extends far enough inside the trap (4) to close the stopper (14) as soon as the initial pumping begins. The assembly pivots around an axis (16) allowing the opening of the shutter (14) when the effect of the counterweight (18) is no longer compensated by the dynamic pressure on the paddle (15) and the pressure difference on either side of the shutter (14). When the flow rate in the siphon conduit falls below a certain threshold, the dynamic pressure on the vane (15) decreases, the upper part of the inverted U-shaped conduit (12) is opened to the atmosphere by the fall of the counterweight (18) which opposes the low pressure at the top of the siphon (4), and the valve (14) is opened. Through this contact with atmospheric pressure, the two columns of water on either side of the siphon (4) fall back under their own weight, preventing any untimely drainage from the downstream reach to the upstream reach. This prevents a pump stoppage from resulting in reverse flow through the siphon and the untimely emptying of the downstream reach (2-A°) into the upstream reach (1-A”). As illustrated in Figure 3 of the bidirectional lock, the flood control siphon will create situations where the water level of a downstream reach (2) may be higher than that of an upstream reach (1). In this case, the locks must be able to maintain watertightness between the reaches and, if necessary, allow vessel traffic. When the two reaches (1) and (2) are at the same level, the downstream gates (31) and (32), around hinges (34) and (35), and the upstream gates (21) and (22), around hinges (24) and (25), move to positions (30) and (33) or (20) and (23) to accommodate the new direction of pressure. The gate halves are operated sequentially to move from the upstream positions (21), (22), (31), (32) to the downstream positions (20), (23), (30), (33) without interfering with each other. The sealing between the left and right gates and the sealing with the vertical walls of the locks is done conventionally by application under the effect of pressure. The solution adopted for the present invention applies to the lower parts of the gates that require the necessary clearance for their rotation. For this purpose, panels similar to panel (27), recessed into the lock floor so as not to obstruct the movement of the gates pivoting above, can also be pressed into position (26) by a jack (29) which, at rest, occupies position (36), and then by hydraulic pressure against the half-gate (23) and against the corresponding part of the lock floor to ensure a watertight seal between the reaches (1) and (2). This panel (26) is stiffened only by a central brace (28) to allow for warping and to compensate for any misalignment between the gate (23) when closed and the floor against which said panel (26) is also pressed. Any imperfections in the seal between the left and right panels generate acceptable functional leakage. Each half of the gate is taken into account by an upstream panel and a downstream panel to be watertight on the three submerged sides: the other half of the gate, the vertical wall of the lock and the bottom of the foundation which constitutes its horizontal part. The following figures and tables illustrate the invention: Figure 1 shows the action of the flood control siphon to lower the level and increase the flow rate. Figure 2 shows the detail of the siphon break vent. Figure 3 shows the principle of the bidirectional lock. The following figures and tables correspond to the method of implementation of the Parisian reach: Figure 4 shows the linearity of heights as a function of flow rate. Figure 5 models the coefficient "C" of the Hazen-Williams formula as a function of the average height of the reach between Port à l'Anglais (PàlA) and Suresnes. Figure 6 shows the acceptable difference between the observed flow rate and the flow rate calculated using the Hazen-Williams formula as a function of the water level at the Austerlitz gauge. Figures 7, 8, and 9 represent the flow rates generated by pumping at Suresnes for water levels at the Austerlitz gauge of 3.3 m (closure of the riverside roads), 5 m (closure of the RER C line), and 6 m (beginning of flooding of the RER C line, 0.1 m below the 2016 flood level), respectively. Table 1 provides the measured flow rates at different water levels. Table 2 provides the interpolated water levels and flow rates during the 1910 flood, with the unsuccessful effect of peaking at 1436 m³ / s in reservoirs without a flood control siphon. Table 3 shows how the flood control siphon used during the 1910 flood would have made peaking at 1750 m³ / s effective. Table 4 also applies to the use of the anti-flood siphon during the 1910 flood and shows how the reduced flow rate in Table 3 makes it possible not to exceed the height at Austerlitz of 5.00 m. Table 5 shows the power required to obtain the flow rate of 1750 m*s as a function of the level at Port à l'Anglais. Detailed description of at least one embodiment The advantages of this invention can be illustrated by the specific case of the Seine in Paris upstream of the Suresnes dam and by some concrete figures, determined from flow rate modeling and the performance of the pumping devices. This invention aims to provide a solution to a threat that has yet to be addressed: a repeat of the 1910 flood. It also allows for the faster emptying of upstream reservoirs and lowers the water level to prevent the almost annual flooding of the riverside roads. Floods in Paris Located at the point where the Seine meanders, a feature evident in its steep slopes, Paris has experienced numerous major floods. Since the flood of 1649, 36 floods have been recorded. -J These are called major floods, meaning those exceeding 6m at the Austerlitz bridge, which occur about once every 10 years. Among them, 16 floods exceed 7m and are considered exceptional. Finally, 3 of them exceeded 8 m, the flood of January 1658 (8.96 m on February 27), that of 1740 and that of 1910. The generating event being the abundance of rain on the basins of the Yonne, the Seine, the Aube and the Marne and the concurrence of their flood waves, it is possible to statistically guarantee that this climatic phenomenon will recur at a time impossible to determine, the qualifier "centennial" of the flood of 1910 does not correspond to its periodicity but to its probability (82 years between 1658 and 1740, 170 years between 1740 and 1910). The current doctrine is based on the leveling of flows by reservoir lakes, on protection by dikes intended to protect Paris and above all the organization to prepare the necessary repairs, for example by optimizing the flooding of all the underground which is inevitable above the 6.20 m level. Everyone knows that the centennial flood will cause considerable damage: in 1910, an additional 1059 million cubic meters (see table 2), beyond the "harmless" flow of 1436 m° / s (closure of the RER "C" at 5m Austerlitz), flowed into the Parisian reach in 39 days (from January 21 to March 6 with two drops to 4.5 m). Table 2 shows that the four artificial lakes (823 million cubic meters) created after 1910 upstream of Paris are insufficient, even with the additional 55 million cubic meters of the "La Bassée" reservoir after 2020.Furthermore, the duration of the flood is of little importance; a few moments are enough to cause damage. While the 1910 flood was estimated at €1.2 billion in 2002, the same flood would generate between €12 and €20 billion in damages today. In 2016 and 2018, "only" 116 and 192 million cubic meters above the "harmless" flow rate of 1436 m³ / s flowed over 5 and 14 days respectively, generating €1.4 billion and €200 million in damages. Insurers estimate that the cost increases exponentially with the height of the flood. As a reminder, retention basins must fulfill two opposing functions: providing sufficient capacity to mitigate flood periods and offering reserves to maintain water flow during periods of drought. To this end, they must be empty around November 1st and then gradually filled until the end of June to support low water levels until the end of October. This low-water support is primarily motivated by public health concerns, secondly by tourism at the reservoir lakes, and lastly by river navigation, which would otherwise be satisfied with the normal water level of 0.84 meters at Austerlitz.This low-water support requires 155 million cubic meters during a dry month to guarantee a flow rate of 80 m³ / s (60 m³ / s from the reservoir and 20 m³ / s from the river). Everyone recognizes that the 885 million cubic meters are insufficient to fulfill these two missions (1 billion cubic meters were requested after the 1910 flood, which materialized with the current 823 million cubic meters plus 55 million cubic meters planned). Presentation of the Paris reach The Seine, 782 km long, rises on the Langres plateau at an altitude of only 446 m. ​​It becomes navigable downstream from Marcilly-sur-Seine (confluence with the Aube), 169 km upstream from Paris, with a drop of 41 m and 19 locks (24 cm / km); 365 km remain before reaching the English Channel, with a drop of 26 m and 6 locks (7 cm / km). As the meanders that appear downstream from Paris demonstrate, the city is characterized by a break in gradient, which this invention would later exploit. The hydraulic networks of the Ardennes, Brie, Langres plateau, and Morvan regions supply Paris. The average flow rate in Paris, at 321 m³ / s, is composed of one-third from the Marne and two-thirds from the Seine. Flood propagation times range from 2 to 11 days. The Oise, whose confluence is located downstream, has a flow rate comparable to that of the Marne. Since our objective is a flood comparable to that of 1910, we can assume that all the networks are saturated and that the dams are flooded. The dams downstream from the Suresnes dam are those of Chatou and Bougival up to the confluence with the Oise, and the Andresy dam downstream of this confluence. The Oise is important because it allows access to significantly higher flow rates: while the probability of a simultaneous flood on the Marne and the Seine is low (1910), the probability of a simultaneous flood for the three Seine-Marne-Oise watersheds can be considered even lower. The Oise network can absorb part of the flow (as in 1910), and the Seine network downstream of its confluence with the Oise is designed for significant flow rates. The Paris reach, upstream of the Saint-Maurice dam on the Marne and the Port-à-l'Anglais dam on the Seine, and downstream of the Suresnes dam, measures 24.990 km.At its Normal Reservoir (RN of 0.84 cm on the Austerlitz scale or 26.76 m IGN 69) the draft (minimum guaranteed water depth) is 3.2 m upstream of the Bir-Hakeim bridge and 3.9 m downstream of this bridge. The heights and flow rates in Table 1, shown in Figure 4, demonstrate that it is not unreasonable to model, within the relevant range of flow rates, the height interpolations as a function of flow rate using the following linear least squares regression parameters: Height = Slope x Flow Rate + Constant Constant slope Port à l'Anglais 3.387 10* 26.696 Austerlitz 3.227 10° 26.191 Suresnes 2.924 10* 24.026 These models will be useful for the interpolations below. Note: all heights correspond to the general leveling of 1969. For the remainder of this text, the Austerlitz heights correspond to the Austerlitz scale, except in sufficiently explicit cases where they refer to the IGN 69 height (offset of 25.92 m), which allows for easier comparison with other traditional references: 8.62 m for the 1910 flood. While heights are often available with high precision, flow rate assessments are more problematic. Thus, the peak flow rate of 2400 m³ / s at Austerlitz (8.62 m on January 28th), usually attributed to the 1910 flood, is, according to more precise sources, 2650 m³ / s. The models presented below have unavoidable limitations; the calculations were primarily conducted to demonstrate the value of accessing more refined models. The dams used are flap gate dams, which can have a vertical retention capacity of up to 7 m and horizontal spans of up to 30 m. It should be noted that these gates retract under the effect of the difference in water level if the chains or hydraulic cylinders that actuate them are no longer under tension or pressure. Consequently, the mechanical modifications required to withstand an opposing force from downstream to upstream do not alter the design principle; only the mechanical resistance of the gates and the dam's foundation needs to be adjusted. Although the available data does not indicate the positioning of the gates over time, in practice the dams are submerged by the flows that interest us and consequently the heights upstream and downstream are identical. The only damage caused by a flow rate of 953 m³ / s (3.3 m on the Austerlitz scale or 29.22 m on the IGN 69 map) is the closure of the riverside roads; at 1436 m³ / s (5 m on the Austerlitz scale or 30.92 m on the IGN 69 map), the RER C line is closed. The basements of the buildings and the RER C line begin to flood at 5.5 m and 6 m on the Austerlitz scale respectively (31.42 m and 31.92 m on the IGN 69 map, 1583 m³ / s and 1731 m³ / s). 'Preliminary note on Hazen-Williams flow modeling:' The empirical law Q = 0.849 CA Rh°** J059 is adapted to the flow rate in channels. Q is the flow rate in m³ / s, C is the roughness coefficient of the channel, 100 for concrete and brick, A is the cross-sectional area of ​​the pipe in m², and Rh is the hydraulic radius (ratio of the cross-section to hmax-hmin wetted perimeter) and J the hydraulic energy gradient ——--— with L the total length of the channel. In accordance with common sense and considerations regarding Froude's number, this flow modeling law reflects the antagonism between slope and cross-sectional area; that is, the same fluid flow rate can be obtained with low gradients and large cross-sectional areas, or with higher gradients and smaller cross-sectional areas. By lowering the water level downstream of the upstream reach through pumping, the flow rate increases by increasing the gradient, up to an extremity beyond which the reduction in cross-sectional area becomes predominant and eventually makes the gain in flow rate negative (see Figures 7 to 9). -10- Hpaia and hsur are respectively the heights in relation to the national levelling of the origins 1969 at the Port à l'Anglais dam and the Suresnes dam. Section A was evaluated based on an observed speed of 1.4 m / s (5 km / h) on February 3, 2018 (see Table 1), hence A = 1100 m (1538 / 1.4). The normal depth of the reservoir is 3.2 m upstream of the Bir Hakeim bridge and 3.9 m downstream of this bridge (average 3.55) at the Austerlitz elevation of 0.84 (average Port à l'Anglais Suresnes of 26.75 m). On February 3, 2018, we obtained an average of 30.35 m (see Table 1 and linear regressions above), i.e., a depth of 3.55 + 30.35 - 26.75 = 7.15 m, from which we deduce a width of 153.85 m = 1100 / 7.15. For other flow rates, A will be considered equal to the product of this width, considered constant, by the average of the anchorage between Port à l'Anglais and Suresnes. The hydraulic radius Rh was calculated using the average height of the mooring between hrai and Hsur. L is worth 24.99 km. The coefficient C corresponds to roughness, 100 in the case of a brick or concrete channel. Since the roughness of the Seine riverbed is unknown, C was determined from the observed flow rate. Figure 5 shows that C decreases from 70 to 50: roughness increases as the river overflows its banks. C was modeled as a function of the average water level between Port à l'Anglais and Suresnes using linear regression with the least squares method. C = -3.0072 x Average (PälA Suresnes) + 153.279 Figure 6 of observed and modeled flow rates as a function of height at Austerlitz shows differences considered acceptable for interpolating flow rates. Preliminary note on the lifting of the flow Available documentation shows that a substantial ship, 320 m long, 37.5 m wide, and with a 14 m draft, traveling at a speed of 11.58 m / s (23.6 knots or 43 km / h), uses a 10 m diameter propeller that generates a thrust of 3 NM (300 tonnes-force) at a rotational speed of 50.4 revolutions per minute and consumes 48 MW (65,000 hp). The rotor disc of this 78 m² rotor is thus traversed by a flow of 909 m³ / s, with a potential lift of 5.38 m (48,000,000 / (909*1000*9.81)). The ocean liner "Normandie" was propelled in 1931 by four 29 MW electric motors; the Queen Mary 2 is currently propelled by four 20 MW electric motors. A TGV develops a power between 10 and 20 MW depending on the model, the network supplying around fifty megawatts (several TGVs per section) -l1- is therefore classic. We can thus compare these flow and height characteristics to those of the Parisian reach. To calculate the mechanical power consumed, the corresponding water level at Suresnes will be estimated from the flow rate of the flood control siphon (using the linear regression above of the natural flow of the downstream reach). This represents the response of the downstream reach to the imposed flow rate and the resulting water level. This level will be compared to the water level generated upstream of the dam by the flood control siphon. P = flow rate x x x g x (natural water level Suresnes - forced water level Suresnes). At this stage, the siphon flow losses and the pump inefficiency are neglected. Effect of the flood control siphon Figures 7, 8 and 9 show the flow rates generated by pumping upstream of the Suresnes dam, as a function of the reduction in height. At 3.3 m on the Austerlitz gauge, when the riverside roads are about to be closed, the calculated flow rate is 953 m³ / s. By drawing water from this section with a power of 6 MW, a flow rate of 965 m³ / s is obtained (an additional 12 m³ / s or 10 m³ / day), and the water level at Suresnes drops by 0.6 m (0.16 m at Austerlitz). Lowering the water level at Suresnes by 1.3 m (0.34 m at Austerlitz) requires 13 MW to restore the initial flow rate. At 5 meters above the Austerlitz gauge, when the RER C line is about to be closed, the calculated flow rate is 1436 m³ / s. By drawing water from this section with a power of 26 MW, a flow rate of 1509 m³ / s is obtained (an additional 73 m³ / s or 6.3 x 10⁶ m³ / day), and a drop in water level at Suresnes of 1.6 m (0.48 m at Austerlitz). Lowering the water level at Suresnes by 3.4 m (0.91 m at Austerlitz) requires 47 MW to restore the initial flow rate. At 6 meters on the Austerlitz gauge, when the RER C line is about to flood, the calculated flow rate is 1731 m³ / s. By pumping water from this section with a power of 45 MW, a flow rate of 1869 m³ / s is obtained (an additional 138 m³ / s or 11.9 x 10³ m³ / day), and a drop in water level of 2.2 meters at Suresnes (0.59 m at Austerlitz). It would be necessary to lower the water level at Suresnes by 4.73 meters (1.27 m at Austerlitz) and consume 78 MW to restore the initial flow rate. These three situations illustrate two fundamental operating modes of the flood control siphon: the "water level reduction and purging" mode and the "water level reduction only" mode, depending on the capacity of the downstream reach to accept a higher flow rate. Both modes protect the RER line and other buildings from flooding. -12- With a downstream reach designed to support a flow rate of up to 1750 m³ / s, the siphon flood control lowers the level by 2.19 m while spending 38 MW (€45k daily). If the downstream reach is not capable of supporting additional flow up to At 1750 m³ / s, the flood control siphon has the advantage of lowering the water level in the upstream reach without increasing the flow rate, as shown by the right-hand side of the curves Figures 7, 8, and 9. The downstream reach will experience exactly the same flow rate as without a siphon. flood protection, but the upstream reach will have been preserved. Analysis of the 1910 flood The 1910 flood followed a very rainy final quarter of 1909: 240 mm, or 70 mm above the 1901-2001 average for the last quarters of that year. 96 mm in July and 105 mm in October 1909 confirm the exceptional nature of that year, which saturated the soil and groundwater. Periods of frost further reduced absorption capacity. Between January 15 and 20, 1910, 40.6 mm, 45 mm, and 80 mm of rain fell on Paris, Châlons-en-Champagne, and Joigny, respectively. Strong warning indicators were already in place. Since the 1910 flood, reservoirs have been built with a capacity of 823 million cubic meters (878 million after the 2020 works). Table 2 shows the level variations in 1910 and deduces the flow rates by interpolation. Let us make the following advantageous assumptions: The works have been completed and offer 860 million cubic meters of reserve (8 millions of cubic meters are considered unusable, particularly for preserve aquatic life). The reservoirs are empty at the start of the flood, despite the need to build up a June reserve of 155 million cubic meters per month of summer drought (60 m³ / s support at low water level). The five tanks are used ideally with great foresight, Through consistency and anticipation, they prevent the flow rate in Paris from exceeding 1436 m³ / s i.e., 5 m at Austerlitz and can generally fill up to 1190 m / s (2650- 1436 m“ / s). As soon as the flow rate in Paris falls below this value, the reservoirs are emptied. to maintain the flow rate at 1436 m³ / s and anticipate a new rainy episode. -13- The propagation delays between the reservoirs and the flow rate in Paris are completely mastered. Despite these ideal assumptions, Table 2 shows that there were still flooding episodes on February 3rd and 13th, and March 2nd, with local maxima of 6.62, 5.61, and 5.58 m respectively. While Paris did not remain flooded for more than a month as in 1910, a single day above 5.5 m caused damage very comparable to a month at that level. The reliability of the calculations rests on the accuracy of the water levels recorded at Austerlitz (which is difficult to dispute) and on the flow rate interpolations (excellent linearity above 1000 m³ / s, see Figure 4); the Hazen-Williams formula, which is more problematic due to the canal's geometry, was not used. These calculations show that a reserve of 878 million cubic meters is insufficient because it offers no margin, despite impeccable implementation, more than 1059 million cubic meters available would have been necessary. Moreover, if this flood, with an excess of 1,057 million cubic meters above 1,436 m / s, could have been managed as effectively, the floods of 2016 and 2018, with "only" 116 and 192 million cubic meters of excess, respectively, would not have caused 1.4 billion and 200 million euros in damage (admittedly, with only 823 million cubic meters of reservoir capacity). It would be interesting to determine why these reservoirs failed to fulfill their role during these two floods. Use of the flood control siphon In order to cope with two months of drought, we will assume that the reservoirs with a capacity of 878 million cubic meters (capacity after the 2020 works) already have 328 (155+155+18) cubic meters at the start of the flood, leaving an exploitable volume of 550 million cubic meters). The logical progression is as follows: 1. A reasonable estimated flow rate of 1750 m* / s is chosen (order of magnitude of the flood) (from 2018). This flow rate must be available upstream of the millrace, and generated within the reach thanks to the anti-flood siphon and suffer it downstream of this reach. 2. This flow rate allows the reservoirs to mitigate the peak flow of the 1910 flood without causing overflow the tanks. 3. At every opportunity, the reservoirs are emptied to maintain this flow rate of 1750 m* / s. -14- 4. So that the Paris reach receives this flow with a level at Port à l'Anglais of 32.37 m, the flood control siphon draws water at Suresnes at 26.95 m with a power of 38 MW. The equivalent elevation on the Austerlitz scale is 5.00 m (drop of 0.93 m), the difference in level on either side of the Suresnes dam is 2.19m. 5. This flow rate must not cause the water level at Port à l'Anglais (1-A°*) to rise above 32.37 m (instead of the 32.62 m that this flow rate of 1750 m* / s generates without an anti-siphon flood; 32.37 m corresponds to a natural flow rate of 1675 m³ / s). 6. The riverbeds and streams between the reservoirs and Paris will have been dredged to that the suction of the anti-flood siphon at 1750 m³ / s lowers the level at Port to The Englishman at 0.25 m (2.19 m in Suresnes and 0.93 m in Austerlitz). In case of a decrease at the top, the flap dam (9) would be slightly raised to reduce the drop to 0.25m. 7. The Seine downstream from Suresnes will have been modified so that this flow rate of 1750 m³ / s does not It does not cause damage. These stretches of road are less urbanized (absence of numerous tunnels under the Seine in Paris), it is desirable to increase their wetting by dredging operations and bringing the dikes into compliance with this flow rate. 8. To limit the rise in water level to 5.00 m instead of 8.63 m in 1910, the energy spent would have had a cost of €1.680 million (at €50 / MWh). Tables 3 and 4, as thus compiled, show that the judicious use of this 550 million cubic meter capacity and the flood control siphon makes it possible to avoid exceeding the 5.00 m level at the Austerlitz gauge and to limit pumping power to 38 MW. This requires the ability to inject up to 866 m³ / s (on January 23rd, compared to the 1190 m³ / s required on January 28th without the flood control siphon). It can be seen that the flood control siphon allows for the rapid emptying of the 200 million cubic meter reservoirs (in 8 days until February 6th) to cope with any new rainfall event. Table 3 assumes a 4-day travel time between the reservoirs and Austerlitz; this travel time is not critical. An important task is to supply the Paris reach at 1750 m° / s without exceeding the level of 32.37 m at Port à l'Anglais.If this were to happen, the flood control siphon could increase its flow rate above 1750 m³ / s in the Paris reach, the level there would fall and cause the level at Port à l'Anglais to fall as well. Table 4 was created using the following formulas: -15- Port à l'Anglais coast = flow rate x 3.387 107+26.696 Suresnes coast = flow rate x 2.924 x 10° + 24.026 Average coastline = (Port à l'Anglais coastline + Suresnes coastline) / 2 Natural slope = (Port à l'Anglais coastline — Suresnes coastline) / 0.2499 Anchorage = Average coast + 3.55 — 26.75 The flow rates in Table 3 are based solely on linear interpolations. Those in Table 4 use the Hazen-Williams formula. The episode from February 28th to March 4th could have been managed by peaking the water levels using retention tanks. However, the flood control siphon was chosen here to demonstrate its effectiveness in managing minor rises. Table S illustrates the sensitivity of the level (1-A””) at Port à l'Anglais to maintain the flow rate of 1750 m* / s while remaining below 5 m. In practice, a sufficient upstream flow rate must be available and regulated by means of the valve (9). Invention with potential for industrial application The use of meteorological and hydrological data is essential to detect the onset of the flow surge in rainfall and watersheds. As soon as conditions allow, the flexibility of the flood control siphon enables the immediate drainage of the available water. The maximum difference in levels between the two reaches is 2.19 m, which poses no difficulty for the classic flap gate dams separating the reaches. Table 4 shows that the flow rate of 1750 ms must be achieved with a power of 38 MW. If the flow in the siphon is accelerated to a speed of 11.58 m / s, the required cross-sectional area is 151 m², for example, 6 siphons with a diameter of 5.66 m, between which the 38 MW are distributed. It may be advantageous to limit the height of the siphon to reduce depressurization by using elliptical cross-sections, for example, 3 m high and 10.75 m wide, to maintain the same cross-sectional area. Each siphon spanning its valve would be equipped with its own propeller pump and motor. All considerations regarding flow rates and power are only preliminary estimates requiring further study; of course, it is essential to incorporate all necessary pressure losses, redundancies, and safety margins. However, within the context of illustrating this patent application, they demonstrate that the technology is already developed to achieve the desired orders of magnitude.

Claims

Appendix C New version of the demands 1. Anti-flood siphon system composed on the one hand of a dam (3) located between an upstream reach (1) and a downstream reach (2); the downstream part of the upstream reach and the upstream part of the downstream reach being in contact with said dam, on the other hand of a siphon consisting of an inverted "U" shaped conduit (4) having one mouth positioned in the upstream reach forming the nozzle (7) and the other mouth positioned in the downstream reach, said inverted "U" shaped conduit spanning the dam (3) separating these two reaches and finally of a propeller pump composed of a motor (5), a propeller (6) associated with the siphon to constitute a siphon-pump assembly having the capacity to locally reduce the level of the downstream part of the upstream reach, increase the slope (1 -A') of this upstream reach, its flow rate, its number of Fronds,and reduce its average level characterized by the mechanical response of the dam (3) inverse to the periods of shutdown of the anti-flood siphon and therefore an effective hydraulic retention in this inverse direction: in normal times the waters of the upstream reach (1-A) are prevented from flowing towards the downstream reach (2-A) and, when the propeller pump is in operation during flood periods to raise the waters of the upstream reach (1-A') to the downstream reach (2-A'), the waters of the downstream reach (2-A') are retained to prevent them from flowing towards the upstream reach (1-A') lower down., 2. Anti-flood siphon system within the meaning of claim 1 characterized by the installation of a “U” between the nozzle (7) and the siphon (4) to redirect the flow upwards and prevent the shaft between the motor (5) and the propeller (6) from passing through the siphon (4).

Images