MWH Soft’s InfoWorks RS 2D solution has been found to perform extremely well in an Environment Agency trial project to model and map the impacts of complex cumulative reservoir dam breaches.
In England and Wales, the Environment Agency is the Competent Authority responsible for ensuring the safety of the 2092 reservoirs that currently fall under the Reservoirs Act 1975.
In addition to this legislation, the Water Act 2003 requires that all reservoir undertakers produce reservoir flood plans for all of their statutory reservoirs. Although this has been a statutory requirement since that date, the floods of 2007 led the Environment Agency to commit to producing Indicative Flood Maps for all of its statutory reservoirs by December 2009.
The RIM project was undertaken in two phases, the first of which involved the development and testing of the draft modeling and mapping specification to be carried forward for use by SFRM consultants during Phase 2 - the main modeling and mapping phase of the study. The draft specification was jointly developed by JBA Consulting and Mott MacDonald. Following development of the draft specification, Atkins and Halcrow were commissioned to undertake a pilot study to test the specification and make recommendations for its improvement prior to the national scale mapping being carried out.
The main aim of the pilot study was to test the draft RIM specification (see box) by applying it, as it stood, to a number of reservoirs to produce flood depth, velocity and hazard maps. A key purpose of this was to provide the Environment Agency with time and cost estimates on a per reservoir basis, to enable them to develop a programme and budget for the national-scale modeling.
Another key aspect of the study was to consider the quality of the study output, in terms of its expected end use and hence to assess the robustness and appropriateness of the methodology provided in the draft specification. To this end, Atkins undertook a series of nine tests to determine whether the methodology was sufficiently rigorous for application on a national scale.
Key tests included examining whether the methodology used to define the breach hydrograph was appropriate or whether alternative breach methods should be used, and whether the effect of buildings should be included within the hydraulic models. One guiding principle of the study was that the maps should provide a credible worst-case scenario, that is, a credible upper limit of what would occur if a reservoir were to breach.
At the time of the pilot study, the draft RIM specification did not have a definitive methodology for the way in which multiple reservoirs within cascade systems should be modeled. Hence a key responsibility of the pilot study was to further develop the methodology for modeling the breach from reservoirs in cascade. Atkins was also required to make recommendations on how the specification might be improved, and to provide feedback on any problems encountered (for instance with data supply, data acquisition and model performance based on the models used).
The Atkins study area was in the north-east of England, and included reservoirs in the Humberside and Yorkshire area. The consultancy was required to model 20 reservoirs, 19 of which had earth-filled embankment dams and one of which had a concrete dam. There were five individual reservoirs and four groups of cascades, which meant that the method for modeling cascades was extremely important to the study.
The Atkins trial
One of the most important aspects of dam breach modeling is the definition of the breach hydrograph. In the draft RIM specification, the Froehlich Equation had been identified to generate breach hydrographs for all dams, which provides different equations for earth embankment and concrete dams. The Froehlich Equation is empirically derived and produces a triangular dam breach hydrograph. For dams in cascade systems, the timing of the start of the breach hydrograph is dependent on the time of arrival of the flood peak from upstream. The ability to use a ‘trigger’ within Infoworks, which activates an inflow when specified hydraulic conditions are met, was invaluable in getting the model timing right within cascade systems.
Atkins used InfoWorks RS 2D for all 20 of the reservoirs that it modeled as part of the RIM pilot study, one example being a cascade of five reservoirs approximately 35km in length from the upstream to the downstream end.
Constructing the models involved processing DEM (digital elevation model) data, either in the form of LIDAR data, SAR data or LIDAR/SAR composite data provided by the Environment Agency. After the data was processed it was imported into InfoWorks RS and an initial estimate of the necessary extent of the model was made.
The model element size was varied for each model and cascade system, but generally the final run was undertaken using the InfoWorks RS 2D defaults for model elements, with a maximum element size of 100m2 even for the larger models. During the entire study, the modelers did not encounter a single model instability or volume balance error within InfoWorks RS, despite encountering depths in excess of 10 metres and velocities of over 5m/s.
The main outputs from the study were a series of maps showing flood depth, velocity and hazards. The maps were created in a GIS by making a thematic map from the standard InfoWorks RS 2D exports of model results. The maps show flood depth and flood velocity for a given location, as well as flood hazard levels.
In Defra’s flood risk methodology, any hazard with a rating greater than 2.5 is considered to be dangerous to everyone including the emergency services. The results from this study often showed hazard values greater than 100. The maps also showed time of initial and peak inundation for the flood.
As well as applying the required specification, the modelers undertook a number of additional tests to check the technical robustness of the method.
One of the tests conducted used the porous wall feature in InfoWorks to simulate the dam crest of each of the reservoirs in a cascade system. The InfoWorks porous wall, which can be assigned a porosity of between 0 and 100%, can be set to partially or fully collapse should user specified hydraulic conditions be met during the simulation. For the purposes of this test, the porous walls representing the dams further down the cascade system were set to collapse should the water level at the dam reach the crest elevation plus half a metre.
When using a method such as this, it is important to ensure that each dam retains the correct volume of water in the model initial conditions, as the lower reservoirs in the cascade system did not have breach hydrographs defined using the Froehlich equation. To ensure that each of the reservoirs retained the correct volume, the Infoworks RS height correction facility was used to modify the DEM. An initial condition was then specified for each reservoir, which was set as the top water level defined in the prescribed form of record or Section 10 report for each reservoir.
In an example case, the furthest upstream reservoir was represented using one of the Froehlich breach hydrographs, with the subsequent reservoirs characterized with collapsing walls to trigger the flood when the relevant hydraulic condition was met. The flood wave progressed downstream very rapidly, breaching the downstream reservoirs in swift succession.
Another test looked at including buildings within the model. Buildings can be included within InfoWorks in three ways: as voids, as areas of increased roughness or as porous polygons. The test undertaken for the RIM study included buildings within the model as voids. Building data was extracted from Ordnance Survey Mastermap data and were applied to the model during the mesh generation process.
Including buildings as voids requires very careful application, as it is possible to generate very small model elements around properties with irregular detail such as bay windows. These small model elements can have a significant negative impact on model run time.
Breach hydrograph method
As discussed, the breach method specified was Froehlich, which is empirically based. Atkins was asked to test whether alternative methods, such as physically-based techniques, might be more appropriate or give a more credible upper limit. InfoWorks RS has a module called HR Breach, which is essentially a physically-based model of how an embankment will breach based on the geotechnical properties of the soil. The module determines not just how the dam would breach, but whether it would breach at all.
In this test, as above, reservoir volume within the DEM was matched to the appropriate volume based on the prescribed form of record or the Section 10 report by modifying the DEM. The outflow hydrograph is very different from the simplistic triangle predicted by the Froehlich method. For the two upper reservoirs, the module represented a two-phased failure with the first peak caused by overtopping before the dam actually failed and the second peak being the outflow after the failure – very different flood mechanisms. There was also a much longer time base for the outflow hydrograph, because the module took into account the time taken for the material to erode and for the breach to develop naturally.
Atkins compared the RIM methodology with the HR Breach methodology to determine which gave the worst case scenario, and found that because HR Breach module models the breach in a more realistic manner, the effect was that the peak flows produced in the downstream valley were much lower. These results were presented to the Environment Agency, and it was decided that the original Froehlich methodology would be used because it would produce the worst case results.
The RIM pilot studies provided essential feedback to the Environment Agency, and Atkins’ study in particular provided feedback regarding how MWH Soft’s InfoWorks RS 2D would perform in producing the national scale mapping.
The feedback that Atkins provided on the draft specification has provided invaluable input to the development of a more robust methodology for cascades, which was only possible because InfoWorks RS 2D made applying the RIM specification relatively straightforward, and provided the right tools for a quick model build. No model stability problems were encountered, allowing the modelers to concentrate on undertaking the tests to determine whether the RIM specification was appropriate. The solution also provided extremely fast export facilities, again making the consultants’ job considerably easier.