When, a few weeks ago, I was asked to present at a joint marketing event with CD Adapco in Aberdeen, a subject wasn’t too hard to choose. How things are changing in the analysis of valves and the world of ASME seemed like an obvious thing to talk about.
High pressure systems, such as valves and pumps, have the potential to cause issues if they aren’t robust enough. But how should you show that a design is going to work, or design towards it working? Especially considering that decoding, understanding and applying what ASME VIII says is a far-from-trivial task.
There is one overriding issue with the design of these components. And that is that at anything like an acceptable operating or test pressure the body of a valve or pump will have small areas that have yielded. So the traditional design approach of ensuring that all parts of a component are kept well away from yield just doesn’t work. And this is an area of technology where simulation has lagged actual practice – because small regions of yielding generally don’t cause component failure. You have to live with them. Understand them. And most vitally, model them.
Stress linearization is a technique which was developed to allow an assessment of the yielding in a structure, and whether this yielding leads to premature component failure. But the big thing about stress linearization is that it is essentially a codified judgement call and requires some degree of creativity in its application. It doesn’t model what actually happens in the failure process; which a proper elasto-plastic analysis would. In fairness stress linearization as an approach dates from a time when a decent linear model of a valve would have been something of an achievement.
But luckily it turns out that ASME does allow the use of an elasto-plastic approach in the assessment of valve bodies, especially in extreme applications. And this has some real advantages; and not just in terms of compliance. By modeling the actual plasticity developing in the solid body you have a picture of what is really happening. And if the growth of the plastic zone is moderate, and doesn’t run away out of control at test pressures, you have a real assessment of the design’s response; which you can use to both show compliance, and understand where you are on the spectrum which ranges from test failure to commercial success.
There is one aspect to this, which, at first glance, seems somewhat counterintuitive, and may, even read like an advert. To show compliance the solution needs to converge at the test pressure (and the rationale behind this is obvious to somebody who spent a couple of days blowing up simple test models). The point at which the solution becomes unstable is the point at which plasticity has spread through the wall. And for that reason compliance is based on the solver converging. Which means that you don’t need to be working with a solver which has convergence issues before this point is reached. Something like Abaqus for example. End of advertorial.
For the presentation, ever aware of the importance of the non-disclosure agreement, we put together a demonstration model which has all the features of a typical high pressure valve. We’ve modeled all the contact interfaces, tied the thing together with 3D solid bolt representations, and used it to show how the yielded material is explicity modelled, rather than having its influence inferred.
So hopefully this marks a step forward in simulation technology in this market space, although it seems that every form of proof of compliance, from long term test and usage arguments to this type of more realistic simulation will all be used for some time to come.
Advanced Simulation
