The development of a high-resolution nested bridge hydraulic model
Date
2023-11-23Embargo Date
2025-11-23
Author
Yu, Haofeng
Metadata
Show full item recordUsage
This item's downloads: 0 (view details)
Abstract
Traditionally, hydraulic modelling of bridges has been undertaken using one dimensional models like HEC-RAS. One-dimensional models simplify river channels
using several cross sections along a river curve and strongly parameterise bridges to
compute water levels and cross-section average flows. Many problems arise because of
the assumptions inherent in one-dimensional models; these give rise to loss of accuracy
and reduced hydrodynamic information. Although there are many drawbacks with one dimensional models, they are still the most widely used in bridge hydraulics. Two dimensional models are much more complex and can be difficult to apply in urban areas
due to the inconsistency in spatial resolution requirements between general urban
domains and channel sections which incorporate bridge crossings. Fortunately, a new
two-dimensional model, entitled MSN (developed at University of Galway) allowing
nesting of different spatial resolutions throughout a domain. Through this approach,
research into two-dimensional treatment of bridge hydraulics can now be considered in
a more realistic manner.
In this research three different approaches to modelling bridge hydraulics have been
coded, in FORTRAN, into MSN; the revised model is entitled MSN_B. This novel
development allows full 2-dimensional analysis of bridge hydraulics at high spatial
resolution, within a larger domain that is modelled at a much lower spatial resolution.
Extensive comparisons were made against the industry standard 1-dimensional HEC RAS model. Results clearly show that along the centre of a channel, underneath a
bridge and downstream in the expansion region, the 1-dimensional model significantly
underestimates velocities. This leads to poor designs for abutments, piers, and aprons.
MSN_B was also compared to a bridge hydraulic model developed within FLUENT, a
CFD model. Results between both models were comparable; thus, MSN_B has the
accuracy similar to a CFD model, but can be used within a model of a large domain,
such as a city, where bridges are nested at very high resolution.
MSN_B was used to assess the characteristics of contraction and expansion zones
around bridges; in particular, where fully expanded flows prevail. It is shown that
recommendations suggested in HEC-RAS are quite different from results of a detailed
MSN_B model around a bridge. Thus, users of 1-dimensional models should take these
new results into account when setting up such models.
MSN_B was originally developed using a Prandtl mixing length turbulence closure
scheme. A minor component of this research was to carry out preliminary
investigations into the significance of turbulence in bridge hydraulics. A k-ε turbulence
closure model was developed within the high-resolution nests of MSN_B; thus, for
computational, efficiency, most of a domain is modelled at low resolution using Prandlt
mixing length closure and regions are a bridge were modelled at high resolution using
k-ε closure. Preliminary results show the k-ε model flow fields are quite different from
flow fields when only Prandlt mixing closure is used.
This research has developed a novel 2-dimensional model for analysing bridge
hydraulics. The model is computationally efficient, as a nesting approach is used,
delivering similar accuracy as a CFD model. Results show that 1-dimensional models
significantly underestimate peak velocities. New recommendations for contraction and
expansion lengths have been developed. It has also been shown that future research
show be carried out on the role of turbulence in bridge hydraulics. A major advantage
of the newly developed MSN_B is that it can be used to fully assess bridge hydraulics
in estuaries; this is not possible with 1-dimenstional models.