The dynamic behaviour of pump gates in the Afsluitdijk

Abstract

As part of the Afsluitdijk project the discharge capacity, which is currently facilitated by two sluice complexes, is increased to cope with future sea level rise. It was decided to realise this by an innovative solution: the pump gate. The steel lifting gates, each containing three pumps, will be implemented in the existing sluice complex in phases, expecting a total of thirteen by the year 2050. This concept provides the possibility of pumping when necessary without hampering the discharge capacity during gravity flow, in which case the gates are lifted. The gates were designed to withstand quasi-static loads. Mainly due to the presence of the pumps, which have a high weight and require limited vibrations, the dynamic behaviour of the gate may lead to more strict design requirements. Two reference designs are investigated: the regular, and flood defence pump gate. The latter is designed to act as part of the primary flood defence, and is therefore significantly more robust. The analysis of the pump gate is limited to three components: the gate structure with its supports, pumps, and fluid. Standard expressions for the hydrodynamic pressures do not apply to the pump gate and surrounding fluid, mainly due to the three-dimensional vibration shape of the gate and the presence of the pumps. General methods or numeric models to quantify vibrations are not readily available for a continuous system with interacting gate, pumps, and fluid. In this thesis, a method is developed to determine the dynamic behaviour of gate-fluid systems confined by sluices. This method is based on a frequency domain semi-analytical coupled modal analysis, able to directly solve the behaviour of gate and fluid for the linearized equations. Several fluid schematizations are found in literature taking surface waves, compressibility, or neither into consideration. The validity of these schematizations was investigated for a wide range of water depths and excitation frequencies. Distinct regions were found in which these physical processes do or do not have an effect on the hydrodynamic mass. The so-called `transition region' is characterised by the absence of both compressibility and surface wave effects. The hydrodynamic mass is therefore frequency-independent in this region and no hydrodynamic damping is present. The response of the pump gate reference designs is quantified by a three-dimensional plate model, based on previously described method. Both designs have considerably higher eigenfrequencies than those corresponding to regular wave excitation. For the Den Oever case, the quasi-static approach therefore suffices when considering wave loads. This is not the case for excitations originating from the pumps, which relate to a wider and higher frequency range. As a consequence of the preliminary design phase, exact pump specifications are not available. Results are therefore based on a pump envelope of possible excitations and presented as risks. These apply to the Den Oever gate designs, but are also relevant to the pump gate concept in general. Three response amplitudes were quantified: the gate's deflection, pump vibration velocity and the resulting fluid pressures. Furthermore, based on the amplification of the static deflection an estimation of the maximum stresses is made. For both designs maximum deflections were less than a millimetre, which is negligible considering the dimensions of the gate. This is explained by the relatively small amplitude of the pump excitation forces compared to the total of static loads. Nevertheless, a significant deflection amplification (dynamic/static) is found. When internal stresses are amplified similarly, a risk of fatigue and even direct failure exists for the regular pump gate design. The more robust flood defence pump gate design reduces stresses to acceptable levels. Vibration velocities of the pumps were compared to ISO limits for non-rotating pump parts. For the regular design at several frequencies these limits were exceeded. The robust design is able to limit these velocities considerably. Most concerning is the magnitude of the pressure fluctuations in the fluid. For the regular design, pressure head amplitudes at the suction side up to 0.5 metre were found. This leads to an increased risk of cavitation, and therewith damage and a reduced efficiency. Furthermore, total head fluctuations over both sides of the pumps can be in the same order as the static pump head. This is expected to lead to an unacceptable reduction in efficiency. The robust flood defence pump gate does reduce these fluctuations over the first range of the investigated excitation frequencies (< 250 rad/s), but for several frequencies leads to a larger response at higher frequencies. %This is likely due to the closer overlap of the structural and fluid eigenfrequencies in that case. %High pump excitation frequencies can therefore be very harmful. The dynamic behaviour of gate and fluid should therefore be considered in further design of the pump gate. Concluding, the dynamic behaviour of the pump gate designs should be considered in further design phases, since several risks were identified. The combination of a robust gate design and limited high frequency pump excitations may lead to an acceptable design.