To Örebro University

In discrete event simulation (DES), the events are random (aleatory) and typically represented by a probability distribution that fits the real phenomena that is studied. The true distributions of event outcomes, which may be multivariate, are often dependent on the values of covariates and this relationship may be complex. Due to difficulties in representing the influence of covariates within DES models, often only the averaged distribution or expected value of the conditional distribution is used. However, this can reduce modelling accuracy and prevent the model from being used to study the influence of covariates. Distributional random forests (DRF) are a machine learning technique for predicting the multivariate conditional distribution of an outcome from the values of covariates using an ensemble of decision trees. In this paper, the benefits of utilizing DRF in DES are explored through comparison with alternative approaches in a model of a powder coating industrial process.

In discrete event simulation (DES) models, stochastic behaviour is modelled by sampling random variates from probability distributions to determine event outcomes. However, the distribution of outcomes for an event from a real system is often dynamic and dependent on the current system state. This paper proposes the use of artificial neural networks (ANN) in DES models to determine the current distribution of each event outcome, conditional on the current model state or input data, from which random variates can then be sampled. This enables more realistic and accurate modelling of stochastic behaviour. An application is in digital twin models that aim to closely mimic a real system by learning from its past behaviour and utilising current data to predict its future. The benefits of the approach introduced in this paper are demonstrated through a realistic DES model of load-haul-dump vehicle operations in a production area of a sublevel caving mine.

Digital twins (DT) form part of the Industry 4.0 revolution within manufacturing and related industries. A DT is a digital model (DM) of a real system that features continuous and automated synchronisation and feedback of optimisations between the real and digital domains. A core technology for predictive capabilities from DT is discrete event simulation (DES). The modelling cycle for developing and analysing DES models is significantly different compared to DM. A DT specific DES modelling cycle is introduced that is evolved from that of DM. The availability of specialised software tools for DT tailored to these differences would benefit industry.

An efficient algorithm is presented for computing exact system and survival signatures of K-terminal reliability in undirected networks with unreliable edges. K-terminal reliability is defined as the probability that a subset K of the network nodes can communicate with each other. Signatures have several advantages over direct reliability calculation such as enabling certain stochastic comparisons of reliability between competing network topology designs, extremely fast repeat computation of network reliability for different edge reliabilities and computation of network reliability when failures of edges are exchangeable but not independent. Existing methods for computation of signatures for K-terminal network reliability require derivation of cut-sets or path-sets which is only feasible for small networks due to the computational expense. The new algorithm utilises binary decision diagrams, boundary set partition sets and simple array operations to efficiently compute signatures through a factorisation of the network edges. The performance and advantages of the algorithm are demonstrated through application to a set of benchmark networks and a sensor network from an underground mine.

Industrial companies rely on hardware and services from external providers to deliver functions that are critical to their operations, increasingly demanding solutions that not only meet technical and availability requirements but are sustainable too. Traditionally, industrial companies choose and purchase hardware and maintenance support to fulfil their functional requirements. An alternative arrangement, known as Functional Product (FP), involves external providers supplying customers with the functionality they require through contracts that specify guaranteed functional availability whilst giving providers freedom to choose and retain ownership of the supplied hardware and services. This paper describes an innovative simulation modelling and optimization approach to quantitatively compare economic and environmental values resulting from transition from traditional to FP arrangements. The approach is demonstrated through the analysis of a scenario involving a hydraulic drive system provider and set of customers in Sweden, with the results exhibiting simultaneous improvement in economic and environmental values at each stage of the transition.

System and survival signatures are important and popular tools for studying and analysing the reliability of systems. However, it is difficult to compute these signatures for systems with complex reliability structure functions and large numbers of components. This paper presents a new algorithm that is able to compute exact signatures for systems that are far more complex than is feasible using existing approaches. This is based on the use of reduced order binary decision diagrams (ROBDDs), multidimensional arrays and the dynamic programming paradigm. Results comparing the computational efficiency of deriving signatures for some example systems (including complex benchmark systems from the literature) using the new algorithm and a comparison enumerative algorithm are presented and demonstrate a significant reduction in computation time and improvement in scalability with increasing system complexity.

Background: Healthcare aims to deliver good patient outcomes. For many clinical procedures there are multiple alternative task sequences that can be performed. These deviations can influence procedure reliability, efficiency of usage of hospital resources and risk to staff and patient safety. Venepuncture is one of the most common invasive procedures in healthcare. Literature of clinical practice shows evidence of wide variability in the procedure order and the duration of each step, which can depend on attributes, such as patient health, sampling method and staff skills.

Objective: To use a computer simulation model based on Petri nets to evaluate the impact on outcomes of commonly practiced deviations from the venepuncture procedure guideline and variations in key dependent variables. The outcomes considered include the probability of successfully obtaining a blood sample and the procedure completion time.

Design: A computer simulation model was constructed using the Petri net technique which mimics the different variations of the venepuncture procedure. Qualitative and quantitative data for the model was collected from the literature and through interviews and questionnaire responses from doctors and phlebotomists. Statistics on the reliability and duration for different variations were then calculated from the model output.

Setting: A digital laboratory to model venepuncture in secondary care.

Results: The model showed that the common practice of applying the tourniquet prior to vein identification and releasing it after sample tubes are filled may result in a ten-fold increase in sample haemolysis, compared to the recommended guideline procedure. Equipment layout on wards and patient vein prominence were identified as the two most important factors influencing time efficiency of blood sample collection.

Conclusions: Petri net computer models were shown to be an effective method for evaluating the success rate and completion time of the venepuncture procedure under alternative task sequences and variations in key dependent variables. The results obtained from the model showed a significant increase in the rate of sample laboratory rejection due to haemolysis when commonly practiced deviations from the guideline procedure were performed. The rate of failure to collect a sample and the mean time for performing the procedure increased significantly for patients with less prominent veins and when the procedure was performed on unfamiliar wards. These results highlight the need for healthcare providers to ensure guidelines are followed when performing venepuncture, equipment layouts are standardised across locations and that the vein prominence of different patient groups is considered when allocating resources for blood sample collection.

A Maintenance Procedure is conducted in order to prevent the failure of a system or to restore the functionality of a failed system. Such a procedure consists of a series of tasks, each of which has a distribution of times to complete and a probability of being performed incorrectly. The inclusion of tests can be used to identify any maintenance errors which have occurred. When an error is identified it can be addressed through a corresponding correction sequence which will have associated costs and add to the maintenance process completion time. A modified FMEA approach has been used to identify the possible tests. By incorporating any selection of tests into the maintenance process it can then analysed using a discrete-event simulation to predict the expected completion time distribution. The choice of tests to perform and when to do them is then made to successfully complete the maintenance objective in the shortest possible time using a genetic algorithm. The methodology is demonstrated by applying it to the repair process for a car braking system. The developed method is suitable for application in abroad range of industries. 

Functional Product (FP) has emerged as a business concept aimed at offering a function or performance, mainly business-to-business applications, on an agreed upon level of availability and cost as well as at providing incitements towards a sustainable growth. Today the literature expanded into various specific approaches and FT solutions measuring sustainability. However, the literature lacks such approaches within the FP context. This explorative paper proposes on a conceptual level a strategy to predict a sustainability impact of an FP in terms of environmental and economic sustainability and optimize the FP configuration. This strategy is based on scenario modelling and simulation driven approach. The practical significance of the proposed strategy lies in its implication to avoid costly trial and error" method performed in the real world and to enable the development of more sustainable products. Through the proposed strategy, it is foreseen that sustainability impact can he quantified and minimised during the FP system development.