This Digital Twinning of Supply Chains article focuses on transformative ideation of transport & logistics value chains. Current practices of Digitization, e-shopping and industry 4.0 have disrupted the market which is embarking for a revision of managed processes, policies and outcomes that may have once served the business well but are now being challenged at the fundamental level.
Any supply chain has complications which creates difficulty in making changes in a piece of the value chain – a balance is required in supply chain orchestration through digital transformation.
An identified knowledgeable approach, best-fit to the challenges caused in switching from an as-is to to-be model is proposed in Digital Twinning where the roughness of data utilized or gathered is proportionate with the problem statement under study.
This orchestration from an As-Is to a To-Be encounters massive data challenges as one moves through several transition phases, each perhaps requiring different modelling methods and progressively finer data tuning in Digital Twinning.
The physical supply chain is not easily changeable. In this Digital Twinning white paper, I have used the term digital twinning in the context of the inherent characteristics of the supply chain that are captured in a digital model. However, such a digital twin varies with modelling method, data visualization, to analytical to optimization or simulation.
Creating out-of-the-box ideas requires a sandbox in SDLC approach for safe experimentation within the digital twins of transformative ideas. Tools in the sandbox have been carefully picked and open to enhancements as it need to be built with bridges. These tools should organize in such a way to deliver interim milestone results and data collection itself is progressive and matched to identically required in the respective digital twin.
I hope that you get more insights from reading this Digital Twinning white paper and that it provides some mapping in your digital supply chain journey.
Digital Supply Chain Hitches
In today’s tech world, DIGITAL is disrupting the way businesses perform across all industries. The mass adoption of digital emerging technologies influences the operations of companies’ logistics and supply chain management.
Smart and interconnected technologies, such as the Global Positioning System, Radio Frequency Identification, cloud computing and sensor devices have changed how businesses interact with their consumers. Customer-centric strategies, innovativeness, flexibility and responsiveness with higher emphasis on fulfilling consumer expectations are the key drivers in this digital era.
Traditional supply chains with linear and long chains may not be sufficient in this digital driven era. Now-a-days businesses need to be dynamic in ratifying the ever-changing trends of consumer demand and shift to a more connected supply network, via digitally interconnected devices and complex platforms to keep pace with digital transformations.
Currently digital supply chain needs to have the capabilities for comprehensive data availability, superior collaboration and seamless communication across value chains.
Here I portray few disruptions in terms of elements and expectations which drives to the need of digital supply chain:
If these disruptive elements not handled properly then it can cause problems and issues in the supply chains, ultimately leading to high operational costs, poor company margins, unacceptable service levels, and low productivity.
These elements enjoin with the business problems of today’s supply chain which is mentioned in below diagram:
Many companies still struggle to make progress with a view to digital supply chain transformation. One of the main reasons for this is that the legacy supply chain and logistics tools/platforms are not able to efficiently address and manage digital supply chain complexities. Therefore, creating a more adaptive and orchestrated platform for assets, business processes, and complex operations has become the imperative.
Data Driven Supply Chain Innovation with Digital Twinning
Internet of Things, Machine Learning and Big Data are at the heart of supply chain digital transformation. It produces enormous data and information that can be in form of structured data such as delivery transactions and warehouse operational data or unstructured data from external resources and social media such as delivery feedbacks. If it managed properly then this data can help generate smarter supply chain and logistics solutions and improve decision making processes.
Hence, many companies are rapidly evolving and investing large amounts of funding and resources in trying to collect and transform data into competitive advantage. However, only collecting (raw data) would not turn the data into business insights. Data processing and analytics, with Artificial Intelligence and Machine Learning technologies are crucial. The raw data needs to be processed into the following steps as shown below:
Supply Chain Understanding and Requirement
This step involves understanding what supply chain aspects are to be improved or identification of the supply chain problems to be addressed before re-shaping the supply chain network. Bottlenecks need to be clearly identified at this stage. To do that, relevant data such as the current supply chain network, supply-demand flow, KPIs are needed.
Data Collection and Acquisition
The next step is to gather these data which are identified at earlier step. This step focuses on data availability & accessibility. Relevant data is collected from different sources, like – Enterprise Resource Planning (ERP) system, sensors, machine generated, social media and external web services. It can be structured or unstructured data, in the format of text, picture, audio or video.
The collected data may be duplicate or with errors. E.g., the same data may be inserted multiple time or timestamp of the data does not match with the fulfilment. It needs to be cleaned before subsequent analysis. This process would include matching record, identifying potential data inaccuracies, making computations for missing data, removing outliers, removing duplications, and formatting the data.
Data Modelling and Algorithm Designing
In this step, mathematical formulas, mathematical / optimization / simulation data models to the supply chain network. It generates insights by identifying relationships among variables, finding patterns from the data, predicting what is likely to happen and optimizing solutions by using what-if scenarios to evaluate transformative strategies for structuring the supply chain network.
Data Communication, Visualization and Business Insights
Once the data is modelled and analyzed using one or more modelling methods and algorithm designs, data along with insights and results from the model can be reported in many formats for communication with the relevant decision makers.
Supply Chain Innovation
Based on the data visualization results of the supply chain, the business owners would be able to take action to transform their network design. It may result in new incremental or radical innovations in the supply chain network. This innovation would be derived from the data and the model used. It would be recorded and updated into the system as new knowledge and insights and can be used for further analysis to derive future innovation.
Supply Chain Orchestration Platform
To address all pain areas of industry by utilizing the extensive supply chain digital twinning orchestration platform. Platform would structure in a way where it equips all parties with proper advocacy in managing changes of goods planning and flow. It integrates supply chain, logistics operations and technologies to strategically shift supply chain resources to create more value and higher returns.
The Digital Twinning platform aims to tackle the main challenges of today’s supply chain that can be summarized as follows:
Supply Chain Transparency
Collaborative data sharing through the whole value chain is still not in usual practice, hence making data available and visible across the supply chain remains as the main challenge. Functional and geographic data silos that do not share information openly, often characterize traditional supply chain.
Usually, vast amount of generated data is stored in a complex and unstructured form that are not system-readable. This leads to less effective performance of the supply chain which are influenced by poor demand planning and management, high operating cost due to excessive inventory, high product return rates and poor SKU service levels due to stock-out.
Supply chain orchestration platform aims to leverage various cutting-edge technologies (i.e. Internet of Things, Big Data Analytics and Machine Learning Algorithms) to provide seamless integration for all processes and activities in the supply chain with secure data sharing which ensures that all stakeholders have the same view of the database to process real-time information automatically. It will permit a supply chain to respond effectively to increase supply-demand, modal choices and demand volatility.
Supply Chain Collaboration
Non-collaborative execution by supply chain stakeholders, particularly in the first and last-mile stage, could result in high costs, low productivity and asset wastage. With limited assets and workforce, supply chain and logistics activities have to be managed in innovative ways to ensure timely order fulfilment.
Collaboration is a strategic term for integrating different technologies, processes, resources, and networks to achieve the optimal operations with an efficient use of whole workforce and assets. One standard approach of supply chain collaboration is delivery consolidation; where data exchange, demand and resource management of more than one stakeholder are synchronous.
Supply chain orchestration platform would enable information sharing across the supply chain to encourage both vertical and horizontal collaboration between the parties of network value chain. Horizontal collaboration for parties having similar logistics requirements can take advantage of potential distribution synergies, such as distribution consolidation and transportation sharing.
For example, using Artificial Intelligence (AI) and machine learning algorithms; the platform would be able to predict the demand fluctuation and fulfilment patterns. These patterns can be matched with patterns from other parties for consolidated deliveries.
Supply Chain Flexibility
Fragmented supply chain network hampers the process and operations flexibility. This nature of supply chains may require significant time and effort to make simple changes. Supply chain orchestration platform would enable real-time planning of inventory and delivery milk runs to dynamically optimize and configure the supply chain to accommodate changing parameterized values such as change of vendor, order quantity, buffer SKUs and lead time.
Dynamic optimization and multi-scenario simulation are the main tools to help networks self-configure to achieve the flexibility. The platform would enable flexibility in determining the distribution network and configuration. With the exploration of multiple scenarios, it would be able to provide more robust solutions that can be evaluated under different kind of criteria.
Supply Chain Intelligence
With the continuous evolving digital transformation and the ever-changing consumer landscape, long chains, functional and geographic data silos, majority of the current supply chains face difficulties to adapt and respond. A discrepancy between production quantity, customer sales forecast and the actual sales may result in lower sales while incurring higher out of stock rate and inventory disposal expenses.
Supply chain orchestration platform consisting of intelligent engines will seek to understand the customers’ demands and reduce the discrepancy between production quantity and customer’s forecast. Using a machine-learning algorithm, it would reveal demand insights and provide suitable forecasting mechanisms in order to maximize revenues, reduce costs / losses / risks within the value chain, increase responsiveness with minimum investment and manpower usage and minimize the mismatch gap of demand-supply.
This can be applied by adopting both the supply chain modules with the current technologies to seed new growth niches, boost its capabilities and translate to a stack of modules. Platform should have an integrated AI powered engine core with data modelling and optimization that provides possibilities for different scenario experimentation, visualization and decision dashboards to give rise to a unique orchestration platform.
The features in the supply chain orchestration platform are divided into three main features, namely: control tower interface, intelligent engine and data configuration and controller.
Visibility & Exception Interface
Supply chain visibility & exception management interface is used to interact with the user and visualize the information and results to the users. The functionalities in this feature can be divided into four groups, namely:
- AS-IS Visualization and Modelling Interface
GPS visualization for supply-demand are the core for this AS-IS interface. It shows the overall supply chain on interactive map view to find out issues, risks and detailed level info for sustain and upgrade the supply chain.
- To-Be (Standard) Modelling Interface
To-Be (Standard) interface would be used to produce optimal scenarios for a particular supply chain, without considering constraints from the industries or companies. E.g., this interface will be used to conduct performing nodes (DCs) of supply chain to identify potential locations for additional warehouses in a particular area or identify risk analysis for a specific supply chain re-structuring.
- To-Be (Practical) Modelling Interface
To-Be (Practical) interface would be used to improve the To-Be (Standard) scenarios for implementation purposes. The scenarios would be generated by considering all practical constraints from the industries and companies, such as limited funding for constructing a new warehouse or land-use regulation for a particular location.
- Dynamic Planning and Monitoring Interface.
This interface can be used for transportation digitalization by providing a dynamic planning and monitoring of the operation supply chain and logistics activities based on the To-Be (Practical) supply chain set-up.
Supply chain orchestration platform would be equipped with intelligent engines to generate scenarios and solutions that will be presented by the visibility & exception interface. Specific engines for supply chain as well as core intelligent engines are integrated in this platform. E.g., supply chain network set-up tool would be selected to determine alternative location for new warehouse, while optimization algorithm in scheduling and routing tool would be selected to produce cost-effective delivery routes.
The integrated engines would create digital twinning of the “physical” supply chain network for evaluating possible improvement scenarios and solutions. The results from these intelligent engines would be sent to and presented in the visibility & exception interface.
Data Configurator and Controller
This feature would capture the data and information from different data source (such as transaction database, social media or sensor data) and store it in the one integrated database. Due to the variability of the data, some data may need to be cleaned before it is used by the intelligent engines.
Let’s explore more scenarios on how the mentioned supply chain orchestration platform is used to tackle a supply chain network problem.
Minimum set of data required for supply chain orchestration platform are:
1. Network Distribution Data
Relevant data on the existing network and distribution (consist of locations of facilities, costs, capacities, available workforce), facility costs, transportation assets, transportation costs and existing routes need to be collected.
2. Transaction Data
Daily transaction data for supplies, demands and delivery schedules are needed. It can be extracted from the ERP system and stored in a particular Central Database Management. Sensor’s data from the vehicles or other logistics assets can also be included to present the actual movement of the goods, vehicles and other assets.
3. Other data
Company policies and considerations are needed to determine the implementation of solution.
The supply chain modelling in this orchestration platform would include the 3 necessary development stages:
1) “As-Is” – Understanding the existing network including issues and improvement areas
2) “To-Be” (Standard) – Identify standard solution for the supply chain network
3) “To-Be” (Practical) – Adjusting T0-Be (Standard) solution based on practical constraints
Using the available data, the As-Is supply chain network can be visualized and modelled. This visualization and modelling will be used to understand the existing situation and identify potential aspects that can be improved in “To-Be (Standard)” and “To-Be (Practical)” model. Examples of this visualization are presented below:
Above diagram visualizes the demands (in orange dots) and demand patterns. The demand can be grouped into several clusters with different central of gravity (in green dots). The dot size represents the number of demands.
Above diagram visualizes the current distribution model (i.e. good flows) from the warehouse (green dots) to the customers (orange dots).
Above diagram visualizes one example of the exiting delivery route to deliver the demands. It applies the milk-run distribution for several customers.
“To-Be (Standard)” Model
The “To-Be” (Standard) model serves as an intermediate model derived from “unconstrained” supply chain network situations. This step would produce an optimized solution based on the model.
To develop the “To-Be (Standard)” model, one should use the Geographic Positioning & Information System (GPIS) with volume of density-based approach which seeks to find the optimum number of storage/freight facilities as well as to define the approximate locations for these facilities. Computations are typically based on minimum transportation costs in consideration of aggregate demand for each customer and product, customer locations and service distances.
In order to build a GPIS model for a particular supply chain network, several inputs are required. These inputs include a list of products, customer locations and the aggregate demand for each customer and product. Typically, the user is further required to preselect a maximum service distance between to-be facilities and customers or a fixed number of to-be facilities. For simplification, GPIS would only consider straight routes between the customers and facilities or the facilities to another facility.
Above diagram shows an exemplary GPIS result. It shows three proposed locations for logistics facilities (in green dot) to serve the demands (orange dots). The GPIS model would build using a simulation software based on two years of operations information on historical demand (by location, amount and time distribution), product flows and costs. The number of facilities can easily be adjusted to analyze the impact on the overall cost-to-serve.
“To-Be (Practical)” Model
The “To-Be” (Practical) model is the final model that includes practical constraints set by the industry or the company itself. This model is an adjustment of the “To-Be” (Standard) model.
The GPIS results may not be able to be implemented directly. It requires adjustment to align with company’s policies and considerations. Hence, a “To-Be (Practical)” model is developed using Network Optimization and simulation by considering the practical implementation constraints.
Network Optimization is used to find the best configuration of a supply chain network structure as well as the flows based upon an objective function, which typically maximizes profits. Considerations for the network optimization are:
1. Transportation cost that is driven by goods flow. The larger the good flow, the higher the transportation cost
2. Fixed cost, the daily cost of operating the distribution hubs. Calculating the daily operating costs per distribution hub, the fixed cost components that drive facility-operating costs were derived from actual cost figures
3. Outbound Processing Cost includes expenses of delivery workforce
4. Inbound Processing Cost includes expenses of warehouse operations
Real Time Scheduling and Monitoring
Standard network configuration derives from network optimization and simulation model which would improve the efficiency of the supply chain and delivery fulfilment. However, supply chains are highly volatile to the exceptions which may have different occurrence frequency and consequences. It would affect the level of service of supply chain network. It would increase delivery lead time and failure to securely deliver the goods.
To anticipate it, a predictive analytics modelling tool to visualize, schedule and monitor delivery schedules for not only the effectiveness but also the robustness of the supply chain network and logistics, fostering the creation of fast response to disruption.
Key Takeaways in Digital Twinning
In this white paper, I address the challenges in digital transformation for supply chain and logistics industry through a supply chain orchestration platform that develop and scale to better efficiency and effectiveness of logistics assets and workforce in digital transformation era.
Consisting of various supply chain modules and techniques, supply chain orchestration platform does foster supply chain transparency, collaboration, flexibility and intelligence to efficiently and effectively cope with the complexities in the digital supply chain.
Finally, the expected outcomes are to achieve an efficient way of analyzing and visualizing data from various sources more time-efficiently schemes to offer optimal prices. This is needed to cope with variable supply and uncertain demand to mitigate risks in the supply chain.
In addition to the various supply chain modules and techniques, this supply chain orchestration platform can be equipped with big data analytics and machine learning techniques utilizing a safeguarding blockchain infrastructure as shown in below diagram. It would enable the companies to take the leap into digital supply chain transformation.
This platform’s ultimate aim has to serve as a digital twinning of the “physical” supply chain network that provides virtuality for evaluating new business scenarios. It enables the user to conduct sandbox testing by changing only a particular aspect in the supply chain by isolating and only changing this aspect to understand its impacts to the overall supply chain.
And at the end, I hope that you (the reader) in turn provide motivation to reshape and collaborate with the same concept to create the supply chain orchestration platform to further enhance the supply chain practices and align with business innovations.
Disclaimer (for map views): This is a conceptualized work of all map mentioned. Geographical Names, projected dots, various stakeholder businesses, cities, and incidents are author’s imagination. Any resemblance to actuals is purely coincidental.