This project is led by Corvus Works in cooperation with their valuable project partners at HR Wallingford:


Unexploded Ordnances (UXOs) and Discarded Military Munitions (DMMs) are found in many coastal regions. They are present in US coastal waters that coincide with former test firing ranges or underwater dump sites and cause a potential risk to personnel as the ammunition or explosive was not detonated. These objects are constantly exposed to environmental influences such as saltwater, which causes corrosion, and waves and currents as well as mobile seabed sediments. Thus, the management and risk assessment of the potentially numerous and diverse object types is a big challenge. For site managers, the location of the objects, along with their type and condition, are of particular interest. Even if these pieces of information were known from surveys, the horizontal and vertical (with respect to the seafloor) locations of particular objects may vary due to continuous environmental influences. Thus, at the least where objects of interest are known to remain on or in the seabed, a scheduled program of monitoring is needed. The project aims to quantify the influence of currents and waves on the mobilization and migration of objects of interest, including the role played by migrating bedforms. This quantification also includes analysis of the (self-)burial and re-exposure of objects. Ultimately, the critical conditions for mobilization need to be identified, along with a probability of occurrence within the area of interest and analysis of how far objects subsequently migrate. This will allow for informed decisions to be made regarding ongoing monitoring, significantly reducing the frequency of repeated and expensive surveys for those locations where the risk of object migration is demonstrated to be low under the prevailing environmental conditions.

Area of Interest

Area of Interest


Environmental data Simulations, using TELEMAC, TOMAWAC and GAIA

We set up a 2D model of the Central Atlantic Coast of Florida, USA, in the openTELEMAC software.  This model encloses the coast between Port St. Lucie and Cape Canaveral.  It includes the Atlantic Ocean waters between the shoreline and ~40 km offshore. Also, it includes the Indian River and Banana River lagoons; as well as the tidal inlets of Port St. Lucie, Fort Pierce and Sebastian connecting those lagoons to the shoreface (Fig. 1).

The model provides a link between global scale hydrodynamics and atmospheric conditions with local currents, waves and sediment processes.  This is achieved by forcing boundary conditions from global datasets (HYCOM and ERA5 models), and simulating its time-space evolution over a finely resolved mesh of Central Atlantic Coast of Florida (Fig. 1).

Simulations proved that the model is able to reproduce measured hydrodynamics of the coast of Fort Pierce throughout 2016, and during Hurricane Matthew (Fig. 2).  These simulations also hindcast the evolution of the sediment bed throughout 2016 (Fig. 3), and are a suitable basis to perform long term UXO mobilization studies through the newly developed UXOmob modelling suite.

An improved burial model DRAMBUIE 2.0 - adding a machine learning approach

The Fig. 1 presents the comparison of three equilibrium burial prediction models and the ANN developed as part of this project. The first comparison presents the results of this comparison for a wave only scenario. As can be seen the modified DRAMBUIE and Voropauev (2003) models have the predict deeper burial depths compared to the Catano-Lopera et al (2007) and the ANN model which are best fit models compared to envelope curves. Further investigation shows that the ANN model has the tendency to approach an S/D value of 1.2 similar to those of the modified DRAMBUIE and Voropauev (2003) models.

The Fig. 2 presents a similar graph as above for a combined wave and current case in which in which the wave height is increase while the current speed is kept constant. The behaviour of the ANN follows more closely that of the modified DRAMBUIE and Voropauev (2003) models. The interesting observation here shows that the ANN predicts a small reduction of the burial depth when small waves are present with a current compared to the current only scenario. This behaviour is also observed widely in the scour around offshore structures where small waves have the tendency to backfill scour holes formed by currents.

Software development UXOmob - combining morphodynamics, scour-burial and mobilisation

The software design depicted in the upper left picture shows a robust architecture that centers on a Core and a Plugin System. The Core is divided into three crucial components: Input, Simulation, and Output, each serving distinct purposes within the software. 

The Input section serves as the gateway to external data sources and defines Interfaces which are implemented in the Plugin System. These Interfaces act as crucial connectors, allowing seamless interaction between the Core and various data sources. This modular approach enhances flexibility and eases the integration of diverse data formats, such as netCDF and csv files. 

The Simulation component interfaces with the Plugin System through three Interfaces: Wave Model, Mobilization Model, and Burial Model. Each of these Interfaces is implemented by specific Computational Models, permitting users to plug in different models as needed. This modular approach streamlines the process of integrating new simulation models, fostering scalability and innovation. 

Multiple Plugins can implement the Output Interface and attach to the Output section simultaneously. Therefore, various output formats, including time series, spatial result storage, or spatiotemporal results can be used in parallel. The software's adaptability to different output formats ensures its versatility in handling diverse user requirements. 

Advantages of this software design include the use of Interfaces, which promote loose coupling between components, making the software easier to maintain and extend. Additionally, the Plugin System enables the dynamic addition of functionality without altering the Core, ensuring scalability and adaptability to evolving needs. This design fosters modularity and reusability, reducing development time and effort while offering the capability to connect with various data sources and output formats. Lastly, the utilization of Data Providers and Data Sinks in the Plugin System allows for efficient management and handling of data, enhancing the software's overall performance and usability. 

A user-friendly graphical interface empowers users to effortlessly select their desired data sources, simulation models and output formats. 

Results - Burial and mobilization of UXO and DMM

The simulations are prepared for six selected objects. The current state of the algorithms, considering only monochromatic waves, is used to compute the burial and mobilization of the 500 lbs General Purpose Bomb. The plots on the right show the state of this object following Hurricane Matthew around Fort Pierce, Florida. The potential mobilization analysis (upper image) shows a lot of activity for the 500 lbs General Purpose Bomb. This is mainly caused by sediment erosion and thus re-exposure of the object. Furthermore the application of monochromatic waves with the wave period being the peak wave period and the wave height being the maximum wave height from the statistical values, tends to strongly overpredict mobilization. In the future, a realistic wave distribution will be considered. The lower image shows the final burial state with regions of complete burial, caused by selfburial and morphodynamics, and regions with lower burial depth. Comparing this result with the morphodynamics, regions of pure self-burial and regions of re-exposure can be found.


The main challenge of the project was to combine different models, covering a huge range of spatial and temporal scales. For this purpose, a new highly modular an parallelized software was developed, which already includes different wave- and burial models as well as a mobilisation model. The software, written in C++, now easily can be extended to consider more different burial and mobilisation models as well as input data from different sources.

Using the example of the full year 2016 incl. Hurricane Matthew at the site of Fort Pierce, it is shown that the model is able to predict burial and mobilisation of different UXO and DMM. As input data, TELEMAC simulations were performed. Furthermore, the already existing DRAMBUIE model was extended and implemented and now considers current and wave action, using an artificial neural network to compute the equilibrium burial depth.

Contact us

Daniel Klembt

Dr.-Ing. Daniel Klembt

Head of the experimental and numerical data acquisition group