Our clients contacted us regarding an emerging issue with a ‘Wiggins Type’ dry-seal gasholder. The piston would not take its full stroke, effectively limiting its capacity. After successfully replicating the failure, confidence in the model was such that it became an inherent test bed for many possible solutions. Ultimately improving the performance of the gasholder, saving money and reducing downtime.
Our client was in the process of developing a steel mill in the industrial area of Santa Cruz near to Rio de Janeiro, Brazil; CSA Siderúrgica do Atlântico (CSA) and contacted us regarding an emerging issue with a gasholder.
The ‘Wiggins Type’ dry-seal gasholder, one of a pair some 58m in diameter and 65m tall, had been designed and installed to function as a storage mechanism for the gases intermittently recovered from the basic oxygen furnace (BOF). These gases would subsequently be used to generate electricity within an on-site power plant.
The dry-seal gasholder comprises a fixed steel cylindrical outer skin and a moving internal piston, of sufficient mass to achieve the desired storage pressure. The outer skin and internal piston are connected and rendered gas-tight using a reinforced elastomeric seal or sealing membrane, that rolls from the inside of the cylinder onto a series of abutment plates on the piston as the height of the piston (volume of gas) changes. The key to the design is that there is no sliding seal interface – which is critical given the massive nature of the structures.
The issue with the gasholder was that, following construction and commissioning, the piston would not travel through its full stroke, but would effectively ‘jam’ at a lower height, thus limiting the gasholder capacity and raising concerns over its longevity.
We have significant experience in the analysis of highly non-linear and dynamic systems and were approached to provide some technical support to our client’s team through a gasholder system analysis using Finite Element Analysis (FEA).
Wherever possible, it is good practice to underpin any analytical task with a number of approaches using different techniques. So, for example, it may be prudent to develop simple hand calculations to understand the likely ‘order of magnitude’ response in the system, to then create simple analytical models or sub-models representing subsets of the system and test their validity before progressing to more complex, all-encompassing models. We applied this ‘layering’ approach to the gasholder system analysis. The ultimate test of any model would be the comparison with the observed behaviour of the actual structure; in this instance the physical manifestation of problems seen during acceptance testing and initial operation.
It was important, when determining the analytical approach, to also consider the likely underlying issue with the gas holder. Given that the ‘Wiggins Type’ design has been in use for many years and that there had been no significant change to the design in its application to this particular installation, we conducted a review of the as-built condition and discovered some out of tolerance geometry. It was therefore critical to our approach that this should be captured in the FEA model.
Starting with an analysis plan, we summarised the bounding load cases and also identified all other critical parameters, such as the data source for the gasholder geometry, material properties, the assessment code (if applicable), software of choice and post-processing requirements.
Given the size and complexity of the structures, a laser point scan of the piston and outer skin was the most practical method to capture the as-built geometry, including any defects. This resulted in a point cloud of many millions of x,y,z coordinates from which it was necessary to weed out any anomalies, for example, birds!
For efficiency, we developed an interpolation macro using ANSYS Parametric Design Language (APDL) to create the outer skin mesh from the point cloud, subsequently varying strake (plate) thicknesses were applied.
The abutment plates, making up the perimeter of the piston, were more complex, so a number of APDL sub-routines were used to position each plate based upon measured data. Lastly, various contact and tension-only elements were incorporated to represent the balance weight system of cables and sheaves and a material model for the sealing membrane was created, respecting its fibre reinforced hyper-elastic structure through a non-linear orthotropic approach.
The ANSYS model of the as-built condition identified that the gas holder was not cylindrical and was also not vertical; through a series of static and dynamic FEA cases it was then demonstrated that the build tolerance contributed to wrinkling of the sealing membrane, which in turn caused the piston to stall before full capacity could be attained.
Through demonstrating that the model replicated current failure modes, sufficient confidence was achieved to use the model as a virtual test bed for a number of corrective actions and ultimately practical interventions that improved the gas holder performance.
- By applying the analytical approach and our and experience, a solution to the problem could be determined more quickly, the plant achieving operational status at an earlier juncture and revenue losses reduced.
- We determined the cause of the failure to perform through a validated analysis
- The analytical model provided a virtual test bed for potential solutions – thus avoiding unnecessary and costly practical interventions