Stability of randomly placed log bed protections

Abstract

Design criteria for the stability of rock filters on river beds (i.e. rock bed protection) are extensively researched and successfully applied in practice. The most common stability criteria are the Izbash and Shields criteria. These methods define a critical flow (Izbash) or parameter (Shields). Rijkswaterstaat (RWS) wants to explore a more sustainable bed protection by using logs. A pilot project is started where logs will be used as bed protection. It is yet unclear if the design and construction approaches for rock bed protections can be used for log bed protections. The most dominant aspect is that logs are cylindrical objects, while rocks are spherical. This means that the design criteria for rock filters might not be directly applicable to log filters. This research aims to verify if the Izbash and Shields criteria for rock can be used for logs to create functional and safe designs.

To achieve this, two experiments are performed at the TU Delft faculty of Civil Engineering & Geosciences. The first experiment, done in a water filled tank, explores the settling behaviour of logs for multiple drop methods. Insight is gained in the settling velocities, horizontal spread and magnus effect (force exerted on a rotating object, e.g. the curve of a football spinning through the air) of logs settling in a water column. Results from these experiments are used in the second experiment. This experiment is done in a 14.3m long flow flume where a log filter is constructed. The log filter is constructed using the drop method that was preferred from the first experiment. The roughness, stability and porous flow of a log filter are investigated. The results are compared with what is known for rock filters.

Tree branches were used as model logs. This was done to be able to correctly scale the results to prototype scale. Primary reason for this was the effect of bark on the roughness of a cylinder. This is difficult to replicate on model scale. Using branches of trees that will be used Saturating the model logs however was more time consuming than initially expected. Attempts were made to accelerate the process but they were futile. One method, using a diaphragm vacuum pump, could not be applied due to lack of resources. For any future research on the topic of tree branches as model logs it is highly recommended to use a vacuum pump to ensure that the maximum density is reached.

From the first experiment it was concluded that large quantities of logs can still be used to create functional log filters. This is a positive result as this will reduce construction time on prototype scale. The method used was a funnel. This method was applied in the flume to create the log filters. By measuring the velocity profile for multiple discharges the roughness of a log filter was measured. This also resulted in an equation of the shear velocity as a function of the discharge. By increasing the discharge step by step, several mobility stages of a log filter were found. This resulted in a dataset that could be directly compared with the Izbash and Shields equations for rock. Higher critical Shields parameters were observed than for rock of the same diameter. The behaviour of a log filter differed from a rock filter in the transition from one stage to another (stable to mobile to transport to failure) did not occur slowly, but almost instantaneous. This behaviour is unwanted because it is difficult to monitor in what stage a log filter is if no changes can be observed between stages. Thus, although applying the equations for rock filters to log filters are conservative, the behaviour of log filters are more sudden and prone to escalation close to their critical thresholds.

Final conclusion is that more research is required to better understand the significance of variables for the settling behaviour and stability. These are water depth, log diameter independent of density and vice versa, log orientation and log length to diameter ratio. This can be done by doing more experiments in a similar fashion where only one variable is changed at a time.

For the application of the design criteria for rock to a log filter with regards to the pilot project, it is recommended to be conservative. Based on these experiments it is safe to assume that when the most unfavourable scenario is used (e.g. low log density, small log diameter, high flow velocities near the bed) a sufficient design is made, especially if the top layer(s) of the filter are placed parallel to the flow direction. Backfilling of the log filter did not increase the stability significantly in this research and is only be beneficial for scour protection.