Testing an inflatable dam3 September 2003
Inflatable dams proved their worth when flooding threatened the Netherlands
THREE inflatable storm surge barriers protect the northwestern province of Overijssel in the Netherlands from flooding. The barriers were designed by Hollandsche Beton-en Waterbouw, a design/build engineering firm headquartered in the Netherlands. Lying at the bottom of the Ramsdiep and Ramsgeul waterways near the city of Kampen, the inflatable dams are automatically deployed when the water reaches 0.5m above normal Amsterdam level (NAP), which is average sea level at the Dutch coast. The average water level at the dam is 0.4m below NAP. Because an inflatable dam of this size had never been built before, the concept had to be validated before it could be built. MSC.Software Professional Services was engaged to investigate the concept, as well as to make design improvement recommendations, using MSC.Marc nonlinear simulation software.
Hans Dries, project manager of HBW, says: 'A prototype is typically a pass/fail situation it either works or it doesn't. When it doesn't work, it's a guess as to exactly why. When it does work, it's difficult to understand whether it was on the cusp of failure or over-engineered. Simulation of the inflatable dam provided the information that allowed us to understand its performance and ensure stresses were within the specified margin of safety. With the inflatable dam, there was no alternative to simulation for validating the design.'
The inflatable dams were designed to work in conjunction with dikes currently undergoing improvement in Overijssel. The area is home to rare and endangered flora and fauna and has dikes, homes and farms with cultural and historical significance, along with unusual locks and pumping stations. Other methods considered for protecting the area could have resulted in flooding an area with unique cultural and historical significance.
The inflatable dams are 75m long, 13m wide and 8.35m high and are made of rubber sheet reinforced with nylon cord. They are automatically deployed and compressors located at each end simultaneously inflate the membrane with air. The top of the inflatable dam is kept above the upstream water level normal, with an internal air pressure between 0.2bar and 0.4bar, with peaks of 0.44bar. As the water level rises, deformation of the membrane increases. The dams are deflated by opening the air valves and pumping the water out. As the dam deflates, the membrane collapses into a sill on the bottom of the river, where rollers move the membrane equally over the width of the sill.
Because approximately 90% of the 2000 inflatable dams worldwide are built by Japan, it has developed the only code that exists for inflatable dams. Based on a 2D cross section, the Japanese code requires a safety factor of eight between initial tensile strength and static load. The safety factors considered include wave loading, stress concentration, ageing, fatigue and water saturation.
Rubber membrane analysis
An FEA simulation using MSC.Marc was performed to calculate the effect of static stresses on the nonlinear material characteristics of the membrane, as well as help understand the performance of the inflatable dam and the inflation/deflation procedure. Scale models in the hydraulics lab were used for correlation with the FEA analysis and for investigating dynamic effects.
Input for the analysis included environmental factors, such as temperature and pressure and the material properties of the fabric resulting from test samples. The 16mm thick membrane fabric was rubber reinforced with nylon cord that weighed 20kg/m2. The membrane for one of the dams weighs in at 33t.
The kinematic and constitutive equations of nonlinear material performance, as measured, are based on small strains the so-called engineering values. In the MSC.Marc analysis the strains are measured in the large strain formulation, therefore to be able to apply the material data in the correct way, strains and stresses had to be converted between both formulations. This particular analysis was very difficult because the membrane is built from a composite of rubber and nylon cord and is quite thin in comparison to its size, resulting in huge displacements that cause instability in the model. Maarten Oudendijk, project manager, MSC.Software, says: 'This is a very thin component, and you have to apply loads and boundary conditions in a way that the analysis remains stable.'
The edges of the membrane are clamped to a sill at the bottom of the river, so it is like a deflated balloon lying on the bottom of the river. However, in the first analysis stresses were higher than acceptable at the clamping points. The critical problem was discovered at the beginning of the 45° angle where the membrane leaves the clamp on the upstream side of the dam. Oudendijk says: 'The main issue was how to clamp the membrane. By clamping it in certain locations, you automatically get very high peak stresses. Moving away from the boundary, peak stresses will vanish because the stresses are redistributed. So we needed a solution for affixing the membrane so it would not break at the points with very high stresses, where you get a kind of domino effect and each cord begins to break.'
Factors related to the excessive stress were identified in the first analysis, enabling recommendations to be made to improve the clamp design. In the first analysis, the model of the membrane had also been clamped in a way that stopped it moving. In practice, when a membrane is clamped and pulled on, it will move. Oudendijk said: 'We knew how to design for lower stresses and our customer knew the limitations for manufacturing the clamp. Together we worked very closely to modify the design in such a way the stresses would be reasonable. With the redesign, we modelled the clamping mechanism like a spring, allowing it to move a little. The simulation determined the force required to start moving the membrane in the clamp and proved the mechanism would work without high stresses.'
In order to prove the accuracy of the simulation, a 1/25th scale model was created in the hydraulics lab. Simulations showed that there were two stable shapes. When the forces were applied to the prototype to create the final shape, just pushing the membrane changed the final shape, from one stable position to the other. However, the loads were lower, which indicated the membrane would not break. Another objective of the simulation was to determine whether stresses after damage caused by a ship's propeller or anchor were acceptable.
Oudendijk says: 'The simulations had a high degree of correlation with the scale model, demonstrating the deformations we identified using simulation were in agreement with reality. At that point, our client and the Government of the Netherlands gained confidence in the analyses we had performed."
On 27 October 2002, the sixth worst storm to hit the Netherlands since 1970 caused the inflatable dam to automatically deploy at about 5:00pm. The gusts of wind reached 120km/hr in the vicinity of the dam, which deployed in less than hour. Within an hour the waters had receded, and, within a further hour, the dam itself had deflated.