1. |
- Abd Alrahman, Yehia, 1986, et al.
(författare)
-
A PO Characterisation ofReconfiguration
- 2022
-
Ingår i: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). - Cham : Springer International Publishing. - 0302-9743 .- 1611-3349.
-
Konferensbidrag (refereegranskat)abstract
- We consider partial order semantics of concurrent systems in which local reconfigurations may have global side effects. That is, local changes happening to an entity may block or unblock events relating to others, namely, events in which the entity does not participate. We show that partial order computations need to capture additional restrictions about event ordering, i.e., restrictions that arise from such reconfigurations. This introduces ambiguity where different partial orders represent exactly the same events with the same participants happening in different orders, thus defeating the purpose of using partial order semantics. To remove this ambiguity, we suggest an extension of partial orders called glued partial orders. We show that glued partial orders capture all possible forced reordering arising from said reconfigurations. Furthermore, we show that computations belonging to different glued partial orders are only different due to non-determinism. We consider channeled transition systems and Petri-nets with inhibiting arcs as examples.
|
|
2. |
- Gutierrez Salas, Mauricio
(författare)
-
Development of Accurate Reduced Order Models in a Simulation Tool for Turbomachinery Aeromechanical Phenomena
- 2018
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Modern gas turbines are still vulnerable to vibrations when operated at certain speeds. This unstable environment can lead to high cycle fatigue (HCF) and damage several of the components inside the turbine. Since engineers are striving to increase the turbines’ efficiency with thinner and more complex blade shapes, these critical speeds will always be present. For these reasons, aeromechanical analyses that is the study of structural and aerodynamic forces need to be assessed with a high level of accuracy. Since this type of analysis are very computational expensive, reduced order models (ROMs) are utilized to decrease the degrees of freedom (DoF) for a faster computation without compromising the accuracy. The present work focuses on cyclic and noncyclic ROMs implemented in an already existing aeroelastic tool, with different characteristics in their condensation and ease of usage depending on the analysis. The AROMA (Aeroelastic Reduced Order Model Analysis) tool has been previously developed to predict the fatigue life of turbomachinery blades with the use of ROMs. The aim of this work has been to improve the tool in terms of accuracy, flexibility and speed, by employing additional reduction methods capable to predict forced responses analysis of large industrial-size models. The understanding of an aeroelastic phenomena would not be complete if mistuning is not considered in the analysis. A mistuned bladed-disk means that all its sectors do not share the same mass and stiffness properties, which in reality this is the case. Mistuning can be addressed as probabilistic, taking into account the manufacturing tolerances and wear of the bladed disk, or it can be assessed as deterministic, also known as intentional mistuning. The latter is achieved to increase the flutter stability by breaking the circumferential traveling waves modes due to energy confinement, and also to have a certain understanding of the forced response amplitude, which helps in designing for worst and best case scenarios. The ROMs that have been incorporated in the AROMA tool are known as the component mode synthesis (CMS) and subset nominal mode (SNM) approaches. The CMS is split into two branches, these are the fixed- and freeinterface methods known as Craig-Bampton (CB) and Craig-Chang (CC), respectively. An intensive study with numerical and experimental validation has been performed for these three reduction methods. The outcome of the study is that each of these methods have their own drawbacks and benefits depending on the aeromechanical analysis problem. The SNM showed that it produces fast computations, with high level of accuracy when the mistuning level is low. On the other hand, a novel and unique approach, Craig-Chang multisubstructuring (CCMS), demonstrated fast computations and high accuracy when the mistuning level is high.
|
|