Luke Thomas Herbert

Precise Quantitative Analysis of Probabilistic Business Process Model and Notation Workflows

We present a framework for modeling and analysis of real-world business workflows. We present a formalized core subset of the business process modeling and notation (BPMN) and then proceed to extend this language with probabilistic nondeterministic branching and general-purpose reward annotations. We present an algorithm for the translation of such models into Markov decision processes (MDP) expressed in the syntax of the PRISM model checker. This enables precise quantitative analysis of business processes for the following properties: transient and steady-state probabilities, the timing, occurrence and ordering of events, reward-based properties, and best- and worst- case scenarios. We develop a simple example of medical workflow and demonstrate the utility of this analysis in accurate provisioning of drug stocks. Finally, we suggest a path to building upon these techniques to cover the entire BPMN language, allow for more complex annotations and ultimately to automatically synthesize workflows by composing predefined subprocesses, in order to achieve a configuration that is optimal for parameters of interest.

Using Dafny, an Automatic Program Verifier

These lecture notes present Dafny, an automated program verification system that is based on the concept of dynamic frames and is capable of producing .NET executables. These notes overview the basic design, Dafny’s history, and summarizes the environment configuration. The key language constructs, and various system limits, are illustrated through the development of a simple Dafny program. Further examples, linked to online demonstrations, illustrate Dafny’s approach to loop invariants, termination, data abstraction, and heap-related specifications.

Restructuring of workflows to minimise errors via stochastic model checking: An automated evolutionary approach

Abstract This paper presents a framework for the automated restructuring of stochastic workflows to reduce the impact of faults. The framework allows for the modelling of workflows by means of a formalised subset of the BPMN workflow language. We extend this modelling formalism to describe faults and incorporate an intention preserving stochastic semantics able to model both probabilistic- and non-deterministic behaviour. Stochastic model checking techniques are employed to generate the state-space of a given workflow. Possible improvements obtained by restructuring are measured by employing the framework׳s capacity for tracking real-valued quantities associated with states and transitions of the workflow. The space of possible restructurings of a workflow is explored by means of an evolutionary algorithm, where the goals for improvement are defined in terms of optimising quantities, typically employed to model resources, associated with a workflow. The approach is fully automated and only the modelling of the production workflows, potential faults and the expression of the goals require manual input. We present the design of a software tool implementing this framework and explore the practical utility of this approach through an industrial case study in which the risk of production failures and their impact are reduced by restructuring the workflow.

Optimisation of BPMN Business Models via Model Checking

We present a framework for the optimisation of business processes modelled in the business process modelling language BPMN, which builds upon earlier work, where we developed a model checking based method for the analysis of BPMN models. We define a structure for expressing optimisation goals for synthesized BPMN components, based on probabilistic computation tree logic and real-valued reward structures of the BPMN model, allowing for the specification of complex quantitative goals. We here present a simple algorithm, inspired by concepts from evolutionary algorithms, which iteratively generates candidate improved processes based on the fittest of the previous generation. The evaluation of the fitness of each candidate in a generation is performed via model checking, detailed in previous work. At each iteration, this allows the determination of the precise numerical evaluation of the performance of a candidate in terms of the specified goals. A discussion of this method’s application, and the degree of optimization which is possible, is illustrated using an example drawn from the healthcare industry.Copyright

Quantitative Analysis of Probabilistic BPMN Workflows

We present a framework for modelling and analysis of real-world business workflows. We present a formalised core subset of the Business Process Modelling and Notation (BPMN) and then proceed to extend this language with probabilistic non-deterministic branching and general-purpose reward annotations. We present an algorithm for the translation of such models into Markov Decision processes expressed in the syntax of the PRISM model checker. This enables analysis of business processes for the following properties: transient and steady-state probabilities, the timing, occurrence and ordering of events, reward-based properties and best- and worst- case scenarios. We develop a simple example of medical workflow and demonstrate the utility of this analysis in accurate provisioning of drug stocks. Finally, we suggest a path to building upon these techniques to cover the entire BPMN language, allow for more complex annotations and ultimately to automatically synthesise workflows by composing predefined sub-processes, in order to achieve a configuration that is optimal for parameters of interest.Copyright

Optimal Sceduling of Stochastic Production Processes Through Model Checking

Modelling and subsequently optimising workflow processes has been a key part of efforts to improve efficiency in production and engineering firms since the beginning of the 20th century (Gilbreth & Gilbreth, 1921).

SBAT. A stochastic BPMN analysis tool

This paper presents SBAT, a tool framework for the modelling and analysis of complex business workflows. SBAT is applied to analyse an example from the Danish baked goods industry.Based upon the Business Process Modelling and Notation (BPMN) language for business process modelling, we describe a formalised variant of this language extended to support the addition of intention preserving stochastic branching and parameterised reward annotations. Building on previous work, we detail the design of SBAT, a software tool which allows for the analysis of BPMN models. Within SBAT, properties of interest are specified using the temporal logic Probabilistic Computation Tree Logic (PCTL) and we employ stochastic model checking, by means of the model checker PRISM, to compute their exact values.We present a simplified example of a distributed stochastic system where we determine a reachability property and the value of associated rewards in states of interest for a real-world example from a case company in the Danish baked goods industry. The developments are presented in a generalised fashion to make them relevant to the general problem of implementing quantitative probabilistic model checking of graph-based process modelling languages.This paper contains three key elements:1. SBAT description.2. Case company description.3. Using SBAT on the case company.The paper concludes by indicating SBAT’s practical applicability and suggests further research directions.Copyright

Addressing production stops in the food industry

This paper investigates the challenges in the food industry which causes the production lines to stop, illustrated by a case study of an SME size company in the baked goods sector in Denmark. The paper proposes key elements this sector needs to be aware of to effectively address production stops, and gives examples of the unique challenges faced by the SME food industry.