Flood Cloud tips
  • 10 Mar 2023
  • 7 Minutes to read
  • Dark

Flood Cloud tips

  • Dark

Article Summary

Undertake test simulations to check configurations

Before running multiple simulations within Flood Cloud, we recommend running one or two test simulations locally to begin with to check that your model is correctly set up. This will ensure that you don’t unnecessarily waste time uploading to the cloud or using Flood Cloud credits on simulations that are going to fail. If you are confident your models are stable, then these local test runs are really only checking that you have all input files in place to enable the model to start-up; you could run these for a short period (and then cancel them).

Flood Cloud Test Function

Flood Cloud includes a test function, enabling you to test your specified simulations are configured correctly and will be able to start running in the cloud. Running a test will save you from wasting Flood Cloud credits on simulations that are configured incorrectly. This test will not ascertain whether simulations will remain stable and run to completion – it is just checking whether all the model files are available.

To use the test function, you must have the relevant modelling software installed locally as this is where the tests will be run (note that test simulations will require an appropriate software licence).

Beware of % characters in filenames

For simulations being submitted to Flood Cloud, ensure the file name does not include the '%' character in the .ief, .xml or .tcf files.

Limit the size of batches

Flood Cloud does not restrict the number of simulations within a single batch. However, we recommend setting up multiple batches with a smaller number of simulations (e.g. 10 simulations) within each, in case a batch fails to complete and to improve efficiency. Note it is currently not possible to specify a definitive limit for a number of simulations in a batch as this will depend the size of the input data files for the included models. Your local internet connection speed will also likely be a factor.

The models in each batch simulation you specify are zipped up into a single compressed file. For TUFLOW models the files for each simulation are kept separate which can lead to the same file (e.g. a ground grid) being uploaded multiple times within a single batch. Note that files shared between multiple Flood Modeller 1D and 2D models in a batch will be recognised and only uploaded once (and then shared between each simulation referencing the file).

Flood Cloud settings checklist

When specifying your batch within the Flood Cloud interface you should consider the following settings:

  • In the 'Options' tab of the 'Settings' pop-up window you can specify a unique name for your batch. This can then be used to identify your batch within the list of active batch runs displayed in the 'History' window (click 'History' button in the Flood Cloud interface to access). If you have left your runs to progress in the Cloud, the 'History' window is used to select and reload the batch during a later Flood Cloud session.
  • The 'Options' tab also allows you to instruct Flood Cloud to prompt for a new identifier for each subsequent new batch you specify. This setting is recommended as it is easy to forget to visit the setup window to define an identifier for each new batch.
  • Your simulation data will be held in the cloud for a period (currently about 1 month) after your batch has completed, and, by default, Flood Cloud will automatically download results as each simulation finishes. Flood Cloud will attempt to download to the specified output folder for each simulation in a batch. You can also set up a default download folder on your local machine for Flood Cloud to utilise if it encounters a problem identifying a specified results folder (settings specified on 'Options' tab of 'Cloud setup' window), e.g. if you are downloading to a different local machine from where the models were uploaded.


Starting batched simulations

If multiple simulations are within a batch, all of the simulations will start at approximately the same time, although there will likely be small variations in start time as different simulations may take more or less time to be uploaded to the cloud.

Once you see the message, “Waiting for simulations” appear on screen, that means the simulations have been submitted, so you can close down Flood Cloud from this point if you wish and the simulations will continue to run.

Identifying simulation problems

The returned data from a cloud simulation will contain all output and log files produced by Flood Modeller and TUFLOW. Selected diagnostic information is streamed back to the user in near-real-time in order to determine if a run is proceeding acceptably, or should be terminated and resubmitted to address an error. Examples of the simulation progress window are shown below. 


If a model is exhibiting stability problems, then this may lead to a high volume of diagnostic information being streamed back to your local interface. In some cases, this can lead to a slowdown in performance (both locally and in the cloud). You do have some control over the amount of information each model sends back. These settings also apply to local batch runs and consequently the settings are accessed from the local batch runner interface.

If you open the Batch Runner interface prior to starting any simulations on the cloud, you can access the batch settings window as shown below (click on 'Settings' button in the Batch Runner toolbar):


You can change how often the Flood Cloud interface will extract run time data from your simulations. The default for this is currently to sample every save interval. However, if you experience a reduction in model performance it is recommended this setting is increased (e.g. to 10).

The 'Limit log output' option is ticked on by default. If ticked then the number of rows stored and shown in the output tab of your cloud simulation is limited to the last n rows, where n is the setting in the 'Show rows' box. You could try using a lower value than the default (100,000) improve simulation performance. The minimum value for this setting is 1.

Downloading results from previous jobs

A feature of Flood Cloud is that you can close both the Flood Cloud interface and Flood Modeller while you have an active batch of cloud simulations running. When you return to the software, you can reload the current batch, or “job”, by clicking the 'History' button on the Flood Cloud toolbar. This will display a new window listing all recent jobs run under the currently logged in account.


The 'History' table has a right-click menu that provides the following options.

To start downloading your results you will first need to select the 'Load' option. This will reload the selected job (i.e. highlighted row in the table) into the main Flood Cloud window. Note that while the batch progress and diagnostics information is downloading there will be a delay which can seem like nothing is happening.

Once your batch run has been loaded, you can select one or multiple results to be downloaded by ticking the appropriate box in the 'Run' column. Then click the 'Download Results' button in the toolbar to start the download process. The destination folder will either be the specified output folder for each simulation or, if this cannot be found, the default download folder specified in Flood Cloud settings (when downloading on a different machine to that which the batch was run from).

TUFLOW model failures with ‘illegal storage access’ error

There is a known bug in the 2016 version of TUFLOW which can lead to problems with TUFLOW in Flood Cloud. This can occur if your TUFLOW tcf file includes the command; “READ GIS WLL” and leads to selected simulations crashing, seemingly randomly. Simulations failing in this way will do so early on in a simulation. The diagnostic message that is returned by these failures will refer to an “illegal storage access” error. In most cases a simulation failing in this way will run through without issue if it is just rerun in a new batch. This bug has been addressed in the 2017 release of TUFLOW.

TUFLOW licence NOT required to run TUFLOW batches

If your batch contains any TUFLOW simulations (e.g. standalone or linked to Flood Modeller 1D) then you no longer need to have a valid TUFLOW licence on your local machine to run these models in Flood Cloud.

Flood Modeller-TUFLOW simulations require TUFLOW relative paths

If you are running a Flood Modeller-TUFLOW linked simulation within Flood Cloud, please ensure that the TUFLOW relative paths are correctly set up, otherwise the model is likely to fail.

Was this article helpful?