Cookies

We use cookies to ensure that we give you the best experience on our website. You can change your cookie settings at any time. Otherwise, we'll assume you're OK to continue.

Research

Research lectures, seminars and events

The events listed in this area are research seminars, workshops and lectures hosted by Durham University departments and research institutes. If you are not a member of the University, but  wish to enquire about attending one of the events please contact the organiser or host department.


 

May 2018
SunMonTueWedThuFriSat
April 2018 June 2018
1 2 3 4 5
6 7 8 9 10 11 12
13 14 15 16 17 18 19
20 21 22 23 24 25 26
27 28 29 30 31

Events for 21 May 2018

Prof Radim Bris: Stochastic renewal process models for exact unavailability quantification of highly reliable systems

2:00pm, CM221

In previous research an original methodology for high-performance reliability computing was developed which enables exact unavailability quantification of a real maintained highly reliable system containing highly reliable components with both preventive and corrective maintenance. Whereas the methodology was developed for systems containing components with exponential lifetime distribution, main objective of this research is generalization of the methodology by applying stochastic alternating renewal process models, so as to be used for unavailability quantification of systems containing arbitrary components without any restrictions on the form of the probability distribution assigned to time to failure and repair duration, i.e. ageing components are allowed. For this purpose a recurrent linear integral equation for point unavailability is derived and proved. This innovative equation is particularly eligible for numerical implementation, because it does not contain any renewal density, i.e. it is more effective for unavailability calculation than the corresponding equation resulting from the traditional alternating renewal process theory, which contains renewal density. The new equation undergoes the process of discretization which results in numeric formula to quantify desired unavailability function. Found component unavailability functions are used to quantify unavailability of a complex maintained system. System is represented by the use of directed acyclic graph, which proved to be very effective system representation to quantify reliability of highly reliable systems.

Contact sunil.chhita@durham.ac.uk for more information about this event.


Prof Coen van Gulijk: Big data risk analysis for the GB railways

2:30pm, CM221

Coen will speak about an exploration into the potential uses of Big Data techniques for implementation in the Safety Sciences. Techniques include text analysis, numeric analysis and the extraction of safety performance indicators from data streams from the GB railways. The work indicates that Data Sciences have a profound effect on safety and safety experts may have to learn data skills for effective safety management in the future.

Contact sunil.chhita@durham.ac.uk for more information about this event.


Dr Shaomin Wu: Higher order Markov processes for the failure process of a repairable system

3:30pm, CM221

Most commonly used models for the failure process of a repairable system have two drawbacks: (1) they assume that the system is composed of one component, and (2) they may contain too many unknown parameters that must be estimated from failure data. However, most real-world systems are multi-component systems and failure data are too sparse to obtain stable estimates for models with many parameters. This necessitates development of new models to overcome the drawbacks. This presentation introduces a higher order Markov process model and investigates its special case, both of which model the failure process of a repairable multi-component system and contain a small number of unknown parameters. We derive a parameter estimation method and compares the performance of the proposed models with nine other models based on artificially generated data and fifteen real-world datasets. The results show that the two new models outperform the nine models, respectively.

Contact sunil.chhita@durham.ac.uk for more information about this event.


Dr Xianzhen Huang : A heuristic survival signature based approach for reliability-redundancy allocation

4:00pm, CM221

Reliability-redundancy allocation can be used to simultaneously determine the reliabilities and the redundancy levels of components to maximize system reliability under design constraints such as those for cost, volume, and weight. Although many efforts have been made to propose efficient and effective methods for solving the optimization model of RRAP, it is still difficult to analyze large systems without considerable computational expense. In this paper, we present a new and efficient approach for reliability redundancy-allocation problems (RRAPs) using the survival signature. The information of the structure function of a system is summarized as survival signature which can be directly applied to calculate the reliability of the system with redundant components in parallel. The RRAP is formulated as an optimization problem with the objective of maximizing system reliability under some constraints. In order to solve the optimization model efficiently, a new adaptive penalty function is proposed to transfer the constraint optimization problem to an unconstraint one. Then a heuristic algorithm called stochastic fractal search is applied to solve the unconstraint optimization. Moreover, the (joint) structural importance is proposed to concretely allocate the redundancy level of each component. Two numerical examples are provided to demonstrate the application of the proposed approach. The results show that the approach is easy to implement in practice and has high computational efficiency.

Contact sunil.chhita@durham.ac.uk for more information about this event.


Dorin Bucur in CM301: Optimal partition problems and the honeycomb conjecture

4:00pm, CM301

In 2005-2007 Burdzy, Caffarelli and Lin, Van den Berg
conjectured in different contexts that the sum (or the maximum) of the
first eigenvalues of the Dirichlet-Laplacian associated to arbitrary
cells partitioning a given domain of the plane, is asymptomatically
minimal on honeycomb structures, when the number of cells goes to
infinity. I will discuss the history of this conjecture, giving the
arguments of Toth and Hales on the classical honeycomb problem, and I
will prove the conjecture (of the maximum) for the Robin-Laplacian
eigenvalues. The results have been obtained with I. Fragala, B.
Velichkov and G. Verzini.

Contact anna.felikson@durham.ac.uk for more information about this event.