(612d) A Transactional Digital Twin for Optimizing Supply Chain Business Processes | AIChE

(612d) A Transactional Digital Twin for Optimizing Supply Chain Business Processes

Authors 

Perez, H. - Presenter, Carnegie Mellon University
Wassick, J., The Dow Chemical Company
Grossmann, I., Carnegie Mellon University
We present an integrated framework for building virtual replicas of the business processes in a supply chain (e.g., order-to-cash and procure-to-pay processes). The framework contains three main modules: process definition, discrete event simulation, and optimization. In the process module, the business process is defined with a graph of tasks and logical operators that govern the flow paths and precedence relations. The graph metadata stores the information necessary to describe the process, such as the resources assigned to each task, task duration probabilistic distributions, and probabilities for alternate path selection. In the simulation module, the business processes are mapped to a queueing network through which requests can flow. The simulation creates value by providing a virtual environment upon which to: 1) test optimization strategies, 2) forecast potential delays in requests based on the current state of the real process and the historical data, 3) identify and mitigate bottlenecks, and 4) provide more accurate fulfillment dates to customers. The optimization module contains heuristics and mixed-integer linear programming (MILP) models adapted from chemical process scheduling (General Precedence [1], State-Task Network [2], [3], and Resource-Task Network [4]). These optimization tools can then interface with the simulation in real-time to assign priorities to the requests in each queue in a feed-back loop. As an integrated simulation and optimization environment, the framework bridges and extends the literature in business process simulation [5], [6] and business process optimization [7]–[9], building upon previous work by the authors that was restricted to only deterministic and static business process scheduling [10].

The proposed framework is developed in the Julia programming language, which brings the advantages of using the powerful libraries available within the Julia environment. These include graph structures via MetaGraphs, discrete event simulation via SimJulia, and mathematical optimization via JuMP. As an integrated platform, the framework does not require the use of APIs to link the simulation and optimization software packages. The platform also streamlines the building and maintenance of digital twins since the discrete event simulation and MILP models can be generated automatically from the process graph. Thus, any modification to the metadata or graph structure, is automatically mapped to the digital twin models. An order-to-cash case study is presented where the digital twin reactively optimizes the priorities of customer orders in a multi-stage transactional network to improve profitability and on-time delivery. A sensitivity study is also shown to assess debottlenecking strategies. Both the case study and sensitivity study showcase the added value of using a digital twin for supply chain business process optimization.

[1] C. A. Méndez and J. Cerdá, “An MILP Continuous-Time Framework for Short-Term Scheduling of Multipurpose Batch Processes Under Different Operation Strategies,” Optimization and Engineering, vol. 4, pp. 7–22, 2003.

[2] E. Kondili, C. C. Pantelides, and R. W. H. Sargent, “A general algorithm for short-term scheduling of batch operations-I. MILP formulation,” Computers and Chemical Engineering, vol. 17, no. 2, pp. 211–227, 1993.

[3] N. Shah, C. C. Pantelides, and R. W. H. Sargent, “A general algorithm for short-term scheduling of batch operations-II. Computational issues,” Computers and Chemical Engineering, vol. 17, no. 2, pp. 229–244, 1993.

[4] C. C. Pantelides, “Unified frameworks for optimal process planning and scheduling,” in Proceedings on the second conference on foundations of computer aided operations, 1994, pp. 253–274.

[5] K. Tumay, “Business process simulation,” in Winter Simulation Conference Proceedings, Dec. 1995, pp. 55–60.

[6] W. M. P. van der Aalst, J. Nakatumba, A. Rozinat, and N. Russell, “Business Process Simulation,” in Handbook on Business Process Management 1, Springer Berlin Heidelberg, 2010, pp. 313–338.

[7] J. Yu and R. Buyya, “A taxonomy of workflow management systems for Grid computing,” Journal of Grid Computing, vol. 3, no. 3–4, pp. 171–200, Sep. 2005.

[8] Z. Cai, X. Li, and J. N. D. Gupta, “Heuristics for Provisioning Services to Workflows in XaaS Clouds,” IEEE Transactions on Services Computing, vol. 9, no. 2, pp. 250–263, Mar. 2016.

[9] P. Hoenisch, D. Schuller, S. Schulte, C. Hochreiner, and S. Dustdar, “Optimization of Complex Elastic Processes,” IEEE Transactions on Services Computing, vol. 9, no. 5, pp. 700–713, Sep. 2016.

[10] H. D. Perez, S. Amaran, E. Erisen, J. M. Wassick, and I. E. Grossmann, “Optimization of Business Transactional Processes in a Digital Supply Chain,” in 30th European Symposium on Computer Aided Chemical Engineering, Sep. 2020, vol. 47, pp. 1159–1164.