Technical Corner
Sensitivity Testing for Hydraulic Models
A hydraulic model is just that: a model. We can’t possibly replicate every feature in the natural world and it is our job as hydraulic modellers to make the best use of the many data sources available to us. At every stage of the model build, we use our judgment to turn this data into a whole host of model parameters which can provide the best representation of reality within the constraints of a computer simulation.
A sensitivity test is a way to try and quantify the impact of some of these modelling decisions. Where a modeller has to make a choice or select a particular parameter, testing that choice allows us to see whether it has a significant impact on results.
We’re often asked to sensitivity test a model as a way of quantifying the uncertainty in the results that arises from uncertainty in the model inputs.
Common sensitivity tests include:
- increase or decrease to the Manning’s n roughness coefficient
- varying the downstream boundary schematisation
- increasing or decreasing the model inflow
The hydrograph above shows the difference in modelled water level when increasing or decreasing the Manning’s n roughness coefficient in the 1D domain.
If there are any other parameters which the modeller has had to select that might have a significant impact on results, it would also be advisable to test these. On a recent project we looked at different ways to represent the outlet losses on a culvert, for example.
Sensitivity testing shouldn’t be a tickbox exercise. The results should steer us towards elements of the model that might require further consideration and analysis or, alternatively, give confidence that a modeller’s choice is not critical and the result is robust. If the model results are sensitive to a certain parameter then it might lead us to advise the inclusion of a freeboard allowance, for example, to accommodate the measure of uncertainty in the result.
Edenvale Young’s modelling experience allows us to build robust models that can help our clients to make well-informed decisions. We still need to be mindful of the limitations and uncertainties associated with those models and sensitivity testing is a key tool in helping us to do that.
Peter Aylett has worked at Edenvale Young since 2006 and more recently became a company Director. He forms part of our technical team and is available to provide training and technical support.
Copyright © 2025 Edenvale Young Associates Ltd. Company No. 5910755
Registered Address: 30 Queen Charlotte Street, Bristol BS1 4HJ, UK