From demand planning and inventory management to production line maintenance and quality control, advanced technologies are being deployed worldwide in factories across diverse industries. Artificial intelligence, machine learning, Internet of Things (IoT, “Industry 4.0”) and proprietary algorithms are advancing the agile, optimized factory of the future.
In its “AI in the Factory of the Future” report, Boston Consulting Group (BCG) reports that the “use of AI can reduce producers’ conversion costs by up to 20%, with up to 70% of the cost reduction resulting from higher workforce productivity.” The application of AI also has been found to help companies better withstand economic downturns. In fact, findings of S&P’s Compustat and Capital IQ, in concert with BCG Henderson Institute, found that 14% of companies leveraging AI increased their sales growth rates and profit margins during the past four global economic downturns. Today, as global supply chains and economies grapple under the pandemic, understanding how to leverage advanced technologies to build future-proof factories has become even more critical.
Industry Giants Lead the Way
As is often the case, leading global brands have been the earliest adopters of advanced technologies to improve their factory operations. Audi, for example, is applying AI for in-line product quality inspections. Similarly, BMW is using the technology for real-time detection of quality deviations on its production line. Machine learning is helping Nokia in its assembly line monitoring, while Canon is using AI, machine learning and predictive analytics to optimize its testing of high-precision machinery parts.
Other ways in which large industry players are applying these technologies include product design optimization, worker safety enhancements, inventory and shipment visibility, supply and demand alignment, improved regulatory compliance, and to advance sustainability goals. There are some core factory processes, universally applicable in businesses of varied sizes and industries, where advanced technologies are paving the way for future gains.
Demand Planning
In a recent survey conducted by the Institute of Business Forecasting, AI was identified as the technology that will have the greatest impact on demand planning over the next seven years. Current supply chain conditions, coupled with volatile global markets, have made demand planning increasingly more difficult. Developing sound forecasts under these and shorter product lifecycles requires more than traditional tools can support. They require robust software solutions that leverage leading- edge technologies capable of addressing multiple dimensions (e.g., product groups, geocentric/regional considerations, customer requirements) in the planning process.
Using graphical representations (i.e., charts and diagrams), a view of the plan is readily accessible to facilitate fast and sound decisions. Simulations of what-if scenarios can test a plan’s resiliency and pave the way for strategic alternatives. If deviations from a plan are detected, the solution’s alert functions intercept them so revisions can be made. Demand planning tools are scalable and adaptable to accommodate large to middle-market and smaller players, all of which benefit from better quality planning, improved service levels, lower inventory levels, and reduced labor planning efforts due to the technology’s self-adapting forecasting functionality.
Inventory Management Optimization
Applying advanced technologies such as AI and machine learning, the 21st century factory is able to maintain the right balance of stock levels, product availability and cash flow. These technologies are being used to provide factories with self-adapting demand forecasts reflecting trends, discrepancies, seasonal peak demands, etc.
Management by exception to address stock-out situations, for example, is supported by graphical displays prompting planners on which actions require prompt attention. Automated order approvals for items forecasted to be in high demand are achieved, thereby enabling inventory planners to focus on high priority tasks. Deploying these inventory management technologies for automated purchasing decisions can reduce a factory’s planning tasks by 25% to 50%, and reduce its inventory by up to 40%, while increasing stock availability.
Contactless Inventory Sampling
Across many industries, an ongoing challenge is maintaining the right inventory levels. The challenge is to support efficient supply chain operations wherein the right level of on-hand inventory is maintained to meet customer requirements, but not excessive wherein heavy carrying costs are incurred. To achieve this goal, inventory accuracy is critical and depends on processes such as annual inventory counting, cycle counting and/or sampling.
In the United States, according to the Internal Revenue Service and generally accepted accounting principles, businesses with physical inventories must periodically conduct an inventory count. Counting methods consist of the annual physical count, a process where a company will halt operations for a few days and count all of its stock keeping units, or cycle counting, where the counting of inventory occurs throughout the years with inventory divided by location product line or inventory type (i.e., more manageable groups). A third and preferred option is statistical inventory sampling.
In inventory sampling, only a small but representative portion of an inventory is counted manually. Then, mathematical- statistical methods are used to extrapolate the documented results of this inventory sampling to the total stock in the warehouse. An analogy would be weighing a single component in order to calculate the weight of multiple components. Inventory sampling has been shown to reduce costs by up to 95% and the effort to conduct an inventory count by up to 99.9%, depending on warehouse variables. Now, thanks to new technologies, inventory sampling can be conducted using a contactless method—especially valuable for containing the spread of a deadly virus like COVID-19.
Contactless inventory applies cloud technology linked to a company’s enterprise resource planning (ERP) system. It relies on a mathematical procedure based on the specific warehouse and which items must be counted as samples. In an ideal application, using a highly efficient sequential testing procedure, the counting process reduces the counting effort to up to 31 positions. Requiring just a few hours on a single day, it does not determine the total value of a warehouse, but rather whether the incorrectly documented stock is within acceptable tolerances. It delivers a sound stock accounting.
While previously, contactless inventory sampling was applicable only in the most automated warehouses, it now can be used for inventory counting in a wide range of warehouses using an infinite number of storage positions. In those cases where sequential testing would not provide enough relevant data, a difference estimation procedure can be used. It calculates the amount of total difference between actual and target stock to validate the accounting. Ideally, this counting procedure is integrated into the daily activities of employees as an advanced permanent inventory counting method which eliminates the need for additional counting staff. It requires a few hours on a single day or up to four weeks.
Inventory sampling, regardless of which procedure is used, does depend on certain criteria, including:
- A warehouse containing at least 1,000 items,
- IT-support inventory management linked to a company’s ERP or warehouse management system,
- Inventory reflecting the Pareto Principle (i.e., 20% of the items comprise 80% of the inventory value), and
- Continued manual counting of high-quality goods, perishable goods and articles with known inventory uncertainties.
Supporting contactless inventory sampling are automated warehouses, drones, measuring devices such as bar codes, magnetic strips and model-driven engineering devices designed to help speed up item grasping without requiring employees to touch objects. Also being used specifically for social distancing during a pandemic is a method where a camera is held by the person conducting the count, who then communicates with another staff member using a video application such as Skype or Zoom. Additionally, a Corona-Tracer, a small, maintenance- free device the size of a matchbox, can be worn by employees to anonymously record approaches by other employees of less than approximately 5 feet. Contacts over the prior two weeks can then be traced so that if an infection occurs, the risk of spread can be contained and communicated, if needed.
Logistics Management
AI, machine learning and predictive analytics are also driving optimizations across the supply chain and its logistics and transportation processes. These agile technologies are evident in optimized preliminary transport capacity planning, real-time transportation scheduling, enhanced time slot management and order processing. Visibility is delivered through digital maps and displays providing real-time data conveying vehicles and order status, as well as key vehicle telematics data such as fuel level, battery voltage and tire pressure.
Yard management systems give organizations complete transparency into their assets’ locations and status which, in turn, promotes increased yard capacity, improved quality control and regulatory compliance. There are also clear cost benefits derived from the reduction in unnecessary asset movement and labor, and associated reductions in detention, demurrage charges and the carbon footprint as it relates to fuel and companywide emissions from a sustainability perspective.
AI Delivering Real-Time Data for the Future- Forward Factories
Through interconnected enterprise software systems, from warehouse management and transportation systems to inventory management and yard management systems, AI-driven real-time data is transforming factories and supply chain processes. None other than the e-commerce pioneer, Amazon, is using AI to optimize its inventory management on an unprecedented scale. AI, machine learning, IoT and other advanced technologies are alleviating the pain points and addressing critical supply chain challenges. They are applying algorithms, which learn by analyzing data and then harnessing the insights gained to deliver predictive models that pave the way for optimized demand, inventory, supply chain and yard management. The outcome is more market-responsive, efficient, profitable and resilient operations.
Justin Newell is chief operating officer of INFORM Software Corp. (Atlanta, Ga.), a provider of AI and optimization software.
