Evaluating supply chain risks from part structures
Enterprise viewpoint
This scenario illustrates how we used the project to explore ways in which early design decisions, such as BoMs and the design of product architectures, impact downstream activities such as the detailed design process which, depending on the design, may be a simple or a complex activity with few or many associated delays. The scenario uses a simulation model to assess these delays so that this knowledge can be taken into account when choosing from different design alternatives.
We established a six step approach to assess product structures (in the form of alternative BoMs for a given part, specified in StrEmbed) using a discrete event process simulation tool (WITNESS) provided by one of the project partners (Lanner Group). This assessment is carried out automatically and quickly: key to being useful at the early design phase due to the volume of options that might be assessed.
The content of this scenario is adapted from a 2022 paper which provides further details of the approach including its implementation.
McKay A, Chittenden R, Hazlehurst T, de Pennington A, Baker R, Waller T. 2022. The derivation and visualisation of supply network risk profiles from product architectures. Systems Engineering. http://dx.doi.org/10.1002/sys.21622
StrEmbed is available from Github here: https://github.com/paddy-r/StrEmbed-6-1, with the following persistent DOI for reference: 10.5281/zenodo.6806818.
Information viewpoint
The example used here is from a case study, a ground support ("parking") trolley, provided by Rolls-Royce plc who were another of the project partners. They wanted to find ways to better understand schedule risks in the design & make supply chains used to produce test products. This is important because delays at this stage often lead to delays in the delivery of products to customers and so reputational damage and increased costs.
BoM transformation is the the process of modifying the assembly structure of a BoM through operations such as addition or removal of parts, grouping of parts into a sub-assembly, flattening (i.e. removing all sub-assemblies to produce a flat list). In this scenario, one of the BoMs is flattened.
Computer support viewpoints
The research established a six step approach to integrate BoMs (specified in StrEmbed) with process simulation tools such as Witness, a discrete event simulation package provided by one of the project partners (Lanner Group). This section includes two computer support viewpoints:
A computational viewpoint that shows the functionality of our approach and so how it delivers the requirements of the Enterprise viewpoint and
A combined engineering/technology viewpoint that shows the system architecture we implemented in terms of key software components.
Computational viewpoint
Step 1: Define product structures and make-buy scenarios
Two product architectures (in the form of Bills of Materials (BoMs) were developed from the parts list for the trolley.
Each had four make-buy scenarios, resulting in eight distinct scenarios. In addition to the BoMs, each scenario included two options: whether or not the Tier 1 parts were made in-house of bought out (in the Tier 1 make-buy column) and whether the capacity in the prime was low (where the prime had capacity to carry out one process at a time) or high (where it could carry out five processes simultaneously). These values can be edited in the Excel file where each scenario is defined.
In the video, StrEmbed is used to modify and export to Excel a flat BoM for the ground support trolley.
First, the user loads the STEP file for an indented BoM, inspects it, then duplicates the BoM into a second assembly. Both BoMs are automatically added to the lattice, a common data structure for the whole project (see Mathematical concepts).
The user then flattens the second BoM, i.e., simplifies the indentation structure to form a flat list, and exports the project to Excel. The lattice view on the right-hand side is difficult to see because of the the number of parts in the parking trolley. This is not unusual and illustrates a need for further work on how best to visualise large structures such as this.
Once opened in Excel, it can be seen that the first sheet is an overview of the whole project, and each BoM that was present in StrEmbed is described in a corresponding Excel sheet containing an indented parts list. The difference in indentation between the first BoM and the flattened BoM is clear.
scenario 4_110722_7.mp4Step 2: Generate supply chain structures from BoMs & make-buy scenarios
Two supply chain structures were generated from the BoMs in Step 1 using macros in Excel. Our Systems Engineering paper provides details of how we do this and, importantly, the assumptions made.
This supply chain structure was derived from the flat BoM structure, and ...
... this supply chain structure was derived from the indented BoM structure.
Step 3: Elaborate each supply chain structure into a supply chain process
For each of the supply chain structures shown in Step 2, a supply chain process was generated. Section 4.3 of our Systems Engineering paper provides details of how we do this and, importantly, the assumptions made. In this resource we show a fragment of the parking trolley BoM: from the parking trolley through to the parts of the wheel assembly.
The following diagram shows a supply chain process derived from an indented BoM where the design and manufacture of the parts on the first tier of the BoM (in this fragment, just the wheel assembly) are outsourced.
The following diagram shows a supply chain process for the same parts but derived from a flat BoM where the design and manufacture of the parts on the first tier of the BoM (in this fragment, just the parts of the wheel assembly but not the wheel assembly itself which does not appear in the flat BoM) are outsourced.
Each structure was stored in an Excel file in preparation for simulation. This file also contains additional data that is needed by the simulation such as nominal timings and rework percentages for different process steps.
Step 4: Translate supply chain process into simulation model
In this project the interface to load and run the simulation is included in the Excel spreadsheet and commands are carried out using VBA. A WITNESS Simulation model is opened that contains an empty modelling canvas and also pre-built modules of all the possible process structure fragments. The data on the process structure is read from the Excel file and used to place modules in the model in the correct places. Data on timings and rework likelihoods is also read in and the model is ready in seconds
Section 4.4 of the Systems Engineering paper provides details of how we translated the process models from Step 3 into Witness simulation models. Each simulation model combined one of the supply chain process models from Step 3 with one of the make-buy scenarios introduced in Step 1: resulting in eight Witness models.
Section 4.4 of our Systems Engineering paper provides details of how we do this.
For example, the supply chain process from Step 3 (duplicated below) ...
... is translated, using the parameters from Step 3 into a Witness following simulation model.
Step 5: Quantify, visualise and experiment with alternative product architectures and make-buy scenarios, and so supply chain risk profiles
To demonstrate the potential value of our approach, each simulation model was run 1000 times automatically (again from the interface within Excel, in a batch mode without displays (for speed).
Excel is ideal for representing the amalgamated data from the many runs and structures to show the timing profiles for design which are the key output from the model. The histograms shown here are results from this process for the eight scenarios introduced in Step 1. In each case, the X-axis is in increments of 40 time units (representing time taken to design and make a parking trolley) and the Y-axis represents the likelihood of a given time taken.
Step 6: Compare supply chain risk profiles with each other
With some human intervention, to align the x-axes, the risk profiles can be compared with each other. This visualisation, generated using Excel to process data from Witness, allows a visual comparison to be made.
It can be seen that there is no obvious best option but the comparison is useful for a number of reasons. For example, identified risks can be managed and prerequisites for apparently good options known. For example, here the best option (all risks low relative to others and no risks in the tail) are based on all design & manufacturing in-house which requires a level of in-house capability (capacity for all required competencies) that may render it infeasible.
Combined engineering/technology viewpoint
Key components of the system architecture used to implement our approach are shown below.
Q: Text to follow
Is it self-explanatory? Perhaps add step numbers?? Can we add links from Witness and Software prototype to further info or downloadable files???
