Supply Chain Digital Transformation and Industry 4.0

Which buzzword seems fancier to you? Digital Supply Chain or Supply Chain (SC) transformation??

We have seen a lot of people touting digital Supply Chain strategy by deploying software. A few questions need to clarify first before diving:

1. What needs to digitalize? What needs to transform? Why does it matter??
2. Will this improve the capability of your Supply Chain in sensing and responding??
3. How to define success when we are in high variability world??

Cash does really matter during inflation time. Inflation is at its highest in two decades, and most companies have the wrong inventory, either in short or in excess. The Consumer Price Index (CPI) rose by 5.2% YoY in February 2023, marking the largest deceleration since April 2020.?

The cash conversion cycle (CCC) is a metric used by businesses to measure how long it takes to convert their investments in inventory and other resources into cash. In other words, it is the length of time it takes for a company to turn its investments into revenue. Let me explain a little more about CCC. CCC is typically calculated by adding the number of days it takes a company to sell its inventory (the days inventory outstanding, or DIO), to the number of days it takes to collect payment from customers (the days sales outstanding, or DSO), and subtracting the number of days it takes to pay suppliers (the days payable outstanding, or DPO).

The formula for calculating the cash conversion cycle is as follows:

CCC = DIO + DSO - DPO

A shorter cash conversion cycle is generally considered better, as it means the company is able to quickly generate cash from its investments. Conversely, a longer cash conversion cycle can indicate that a company is facing liquidity issues or inefficient operations. As such, companies often try to improve their cash conversion cycle by implementing measures such as better inventory management, faster collection of receivables, and negotiating better payment terms with suppliers. The goal of SC digital transformation needs to link with the finance index, such as CCC. The first item to manage is the inventory.?

To understand inventory, one needs to know the type of inventory and how inventory is built in which SC node.?

The diagram shows how demand and supply collaborate in business at different levels: 1. All demand planning and supply/capacity planning link to the strategic plan. 2. Priority planning is the result after considering the constraints. 3. The process is iterative from a high level and rough state and keeps refining and breaking down into component levels as time goes on. 4. The objective is to find the right balance tradeoffs between supply, demand, finance, and business goal. 5. The outputs are purchase orders to vendors for specific parts, quantities, and required delivery dates and/or shop floor production schedules.?

In the book, Manufacturing Planning and Control System (MPC), edited by Vollmann, Berry, and Whybark in 1997, they named the following system, MPC, which has a very similar concept. MPC provides information and support which enables managers to efficiently direct the flow of material, manage the utilization of people and equipment, and respond to customer requirements by utilizing the capacity of suppliers, internal facilities, and in some cases customers.?????

MPC has three parts. The front end helps in setting the overall direction of a manufacturing company. This is where the overall game plan was built, discussed, and aligned. In another word, sales and marketing strategy has to be aligned with the operational resources. Without a seamless integration, the result will be a loss of sales and excess/slow-moving inventory. Demand has two forms, forecast and orders. Manufacturing company needs to deal with the natural volatility of demand and keep competitiveness. Demand management provides the input to sales and operations planning (S&OP) so that the overall game plan can take into account any changes that might be there on the other end. Another key input is what kind of resources we have at the high level, at the plant level, and at the organization level, how much capacity do we have there has to be a match between what kind of demand we are managing and how many resources we have and what additional resources do we need to prepare ahead. S&OP results in a plan, with priorities and constraints so master production scheduling (MPS) is a period-by-period plan for the finished product and product options we have to make to fulfill customer requirements (It may not be appropriate to plan every single finished product that can be made and option level can be much more manageable).?

The middle part is the engine. It handles most of the detailed computation in how many parts have to be made, and how many parts and components have to be brought together for the product options and that kind of calculation requires a large computation within the planning horizon. Detailed material planning is where all the parts, components, and sub-components go into the finished product planned. Therefore, bills of material (BOM) is the key input. Detailed material planning is also called MRP (Material requirement planning). Detailed capacity planning is done at a much finer level. We need to understand exactly how the different labor resources, material resources, and machine resources that we have will help us manufacture these parts into finished products in the required quantity and in time.?

The back end is the execution system that provides information, such as how various activities should be scheduled on the shop floor of the manufacturing company and/or how they should be scheduled at the supplier site.?

MPC is top-down planning but the execution of this would be the basis to fulfill all the requirements. S&OP reflects the overall strategy to meet the customer demand of the future and provides long-term support activities to ensure that the appropriate amount of capacity (Mix of human resources capabilities, technology, and geographic locations) and resources including internal manufacturing resources and external suppliers’ available capacity. Usually, the planning horizon is more than one year and can be extended to 5 years, as aware of long-term planning.?

Intermediate-term planning of MPC supports the activities happening on a monthly basis (It usually covers from half-year to a year depending on the longest supply lead time). The fundamental aspect that we have to focus on now is how we match the supply with the demand. At this level, we have to know what to manufacture (The right mix of product options/configuration, and the right volume in a certain period). Logistics and warehouses are also needed to plan ahead. We will form an inventory plan and place the appropriate inventory levels in the supply pipeline. The periodic plan is generated and requires continuous review and adaptation to change. We will also inform customers how we (will) have satisfied their requirements, such as when to deliver what in which volume to where. When in the outsourced model, you will communicate the requirements to suppliers and get back the supply information back as a customer.?

Short-term planning of MPC supports the activities such as detailed scheduling of resources for day-by-day operation. People and machine work on the RIGHT thing. MPC tracks the use of resources, monitors execution results (Daily throughput, yield, rework, inventory level in RM/WIP/FG), and provides the data to the line managers, customers, and suppliers.???

Till now, I have introduced the MPC system (One of the core modules of ERP); the reason why inventory is produced and corresponds to the type of inventory. ERP can’t work alone. It may work with PDM, WMS, and MES as shown in the following flow.

Some companies have even more complex setups such as below while combining the utilization of AI/ML and IoT.

Before answering the questions at the beginning of the article and proceeding to share a clear path to SC digital transformation, I’d like to explain the concept of Industry 4.0. In 2011, Germany pursued the Industry 4.0 initiative as a critical research and development direction to create more flexible production systems. The initiative aims to address the challenge of increasing production complexity caused by various factors, such as rising global competition, product diversity, and the need for customization to meet customer demands. To achieve this goal, Industry 4.0 envisions an interconnected system that combines information technologies with the production process to enable intelligent manufacturing. However, transforming current production systems to meet these objectives will be a lengthy and challenging process that requires a well-structured migration path. I will share my experience on how to be successful in this long journey at the end of the article.?

The Industry 4.0 initiative aims to develop flexible production systems that can handle production complexity and product life cycles by improving vertical and horizontal integration of production process participants. This integration requires a global connection of information technologies (IT) with the production process, enabling cross-domain data access and sharing of domain knowledge across all relevant production domains. To achieve horizontal and vertical integration of production components, a common understanding of the process is crucial. The Reference Architecture Model for Industry 4.0 (RAMI 4.0) has been published to integrate elements of Industry 4.0 into a three-dimensional layer model, giving formal definitions of their vertical and horizontal integration and product life-cycle engineering.

Digital transformation of production systems requires a coherent migration path, and an approach has been proposed to describe possible stages of the transformation of legacy production systems towards smart factories in the sense of Industry 4.0. Meanwhile, the change of existing running systems needs to stay minimized. Here is the interpretation of RAMI 4.0.?

6 layers represent the IT structure of a production system and the asset role.?

“Life cycle and Value stream” axis is the lifetime of an asset and value-added process based on IEC 62890. “Hierarchy level” axis shows the hierarchical levels of a manufacturing system based on IEC 62246 and 61512 standards.??

“Asset” layer can be seen as the physical world, such as machines, equipment, spares, etc.?

“Integration” layer is a transition from “Asset” layer to information, which means it contains the properties and the process-related functions that asset can work properly as it is defined/designed. Each event in the real work can be found an event in the virtual world in a very short time or real-time by being reported to the Integration layer.?

“Communication” layer describes the connection of the asset to information and other assets.?

“Information” layer contains all the data used, generated, or modified by the technical functionality of the asset. The fundamental requirement of this layer is the consistent integration of different data.?

“Functional” layer can be seen as the twin of the physical asset. It contains a digital description of the technical functions of the asset and its role in the production system as well as fits in the platform for horizontal integration of different functions of all other assets. ERP is in Functional layer.?

“Business” layer orchestrates the activities in Functional layer to assure the integrity of the business model, value chain, and all business processes resulting in a strategic enterprise view.?

The journey to Industry 4.0 is a long-lasting evolving process and it almost changes the fundamental of the whole business. The change needs to be planned and implemented in a strategic way to assure a positive effect on results; small results such as traceability, and transparency, large results such as profitability, growth, cycle time reduction, increased revenue, etc. Schuh G, Anderl R, Gausemeier J, ten Hompel M, Wahlster W provided Industrie 4.0 Maturity Index as shown in Fig below. The path to Industry 4.0 requires 6 sequential steps. The competencies need to be built up step by step from left to right. The first one is Computerization. It is an isolated use of a machine or equipment, for example, using a milling machine with computerized numerical control (CNC). Repetitive activities can be done with higher efficiency/ shorter cycle time and higher effectiveness/better quality. Connectivity is the capability of machines or equipment to communicate with each other. The communication of API from OT (Operational technologies) to business IT needs to implement. For example, handling robots in PCB assembly line can communicate with each other and report to UPH (Unit per hour) to MES. Both Computerization and Connectivity are under Digitization. Next, Visibility takes place; a digital model of the enterprise is real-time available. Dr. Michael Grieves was the first one to apply the concept of “Digital twins” to manufacturing in 2002 and formally announced the concept. You may see the digital twins as a virtual model in the digital world designed to reflect a physical object accurately. For example, a large high-sensitive semiconductor equipment transportation requires to be shipped by 40” container with various sensors related to vital areas of transportation reliability, such as shock meters, vibration meters, humidity meters, and temperature meters, in different areas of the container. These sensors produce data about different aspects of the physical transportation reliability performance. This data is relayed to a processing system and applied to the digital copy. The virtual model is used to run any kind of simulations and analyze the performance issue, then generate possible improvements. All the valuable results and insights can be applied to physical objects. By applying PLM (Product lifecycle management) and digital twins, the company will be able to shorter the new product development cycle time and cost, lower the maintenance cost and increase the service level, and create synergy while applying AI, IoT, cloud, BI, and 5G. The following step, Transparency, all the visible data is aggregated and simulation/analysis has performed. For example, edge computing and image recognition based on AI is used to detect errors or mismatch of data, such as appearance defect, or inventory miscount. Then, the future event can be predicted. We are moving from Transparency stage to Predictive capacity stage. Using the defect data combined with the frequency in a period of time for the working equipment. One will be able to predict the possible equipment failure and get spares ready to maintain (Not recovery) before shut down or more defects are generated. We provide the target influence by monitoring the condition and generating the predictive model before the system behaves unexpectedly. The last step is Adaptability. The system is capable of self-optimizing (Autonomous) based on the data collected, through the calculation of a mathematical model/algorithm, and makes the decision on its own.?

A key influencer of inventory is forecast. I have written four documents about it:?

1. https://www.dhirubhai.net/pulse/forecast-1-statistical-analysis-lo-ping-chang/
2. https://www.dhirubhai.net/pulse/forecast2-professional-judgement-lo-ping-chang/
3. https://www.dhirubhai.net/pulse/forecast-3-strategy-lo-ping-chang/
4. https://www.dhirubhai.net/pulse/forecast-4-demand-planning-robin-chang/

Nonetheless, AI/ML, BI, and cloud provide more imagination in this topic. If a forecast model is not only generated based on historical data (Sales, inventory, call volume) but also considers:

1. Related data from the market: Price band, promotion happening, stocks on the shelf, weather data, promotion events, etc.?
2. Metadata: Color of the item, the target selling city, brand, etc.???

I wonder how much improvement can be made based on the model. I believe GCPs (Global cloud providers) all have this kind of service. You can’t beat Giant but you can stand on their shoulder to build value for your customer.??

Another key agenda is sales. I see many companies failing to implement CRM system, like Salesforce. The reasons for the failure were similar to one another - the expectation that all employees would adhere to the system's rules, which was not the case. The company needs to focus on optimizing its processes comprehensively by engaging experienced sales consultants to help define the appropriate processes for each sales stage. Only after the processes were defined did they implement CRM, ensuring that employees knew what actions to take in response to different processes. With CRM in place, sales operation becomes simpler. Define several selling stages and let the salesperson know exactly which businesses and projects require which resources (With standardized operations, sales teams have to reach certain stages and obtain specific information before they can request the development team to perform the tasks), instead of in the past where sales would directly pass on custom requests from customers to the development team, resulting in overwhelming workloads for the development team with little success in closing deals. Furthermore, CRM also provides the data such as the average sales cycle for new customers is xx days, while for old customers, it's within xx days. As a result, each salesperson can now rely on the system to calculate potential sales volumes ahead based on the average sales cycle, rather than relying on intuition, allowing for a more accurate allocation of logistical support.

Based on my experience in SC digital transformation, a few success factors are suggested: 1. Define a proper goal that links to business: Reduce cycle time, reduce inventory, increase the cash conversion cycle, increase the responsiveness of ATP (Available to promise), and improve the on-time-delivery rate (OTD) are good examples. 2. People who have cross-functional skills are extremely valuable: You will need an experienced team to know the organization, know the business goals and know the processes. You may also need external consultants to drive the changes from outside in and irritates new thoughts. 3. ? Processes: Review the processes and see if they are well-defined, documented, and updated. Rethinking processes based on technology advancements to drive improved business capabilities. 4. System: Whether the system you consider is at the Enterprise level, work center level, or device level, the system needs to be adaptive to your process. Meanwhile, people need to know the logic of the new system and understand why it can help. ?

Appendix A: Myth (Q&A):

1. Is Digitalization equal to paperless? Digitalization and going paperless are often used interchangeably, but they are not exactly the same concept. While they share similarities, there are key differences between the two. Digitalization refers to the process of converting analog information into digital form. It involves using digital technology to capture, store, process, and manage data in electronic formats. This can include scanning physical documents into digital copies, using digital tools to create and edit documents, and storing data electronically in databases or cloud storage. On the other hand, going paperless refers specifically to reducing or eliminating the use of paper in favor of digital alternatives. This could involve practices such as sending and receiving electronic documents instead of printing and mailing physical copies, using digital signatures instead of physical ones, and using electronic communication tools for collaboration and information sharing.
2. Automation means hands-free. Is that a goal for SC digital transformation? I think automation is NOT the major object to achieve because from all the successful programs I have managed the win comes from reducing none value-added work and streamlining the process flow.?
3. Does real-time matter? Real-time data means the data is not yet processed and we are not sure if the data contains noise or outliner when not enough samples can be obtained. My suggestion is to define data with time sensitivity. Process the most important and time-sensitive data to prevent latency.?
4. Will digitalization improve the data inconsistency? Unfortunately most of people expect the inconsistent data problem will be gone after the implementation of an expensive system. However, it is irrelevant to whatever system you use. You must clean the old data first before feeding any new data. Second, new data should follow the same format, same template, single input with sanity check.Tableau shares the quality data having 5 characteristics: Validity, Accuracy, Completeness, Consistency, and Uniformity (https://www.tableau.com/learn/articles/what-is-data-cleaning#components-quality-data) which can be a good reference when you define/refine your process input and output; moreover, quality data can be generated.