Introduction

“Composability” is key for Manufacturing Execution Systems (MES) to help manufacturers roll out across an entire fleet of factories on time, on schedule and with cost efficiency.  Until now, this ability was largely unachievable.

Manufacturing company CXOs manage stringent financial targets as they juggle various challenges. The supply chain volatility, skilled labor shortages, factory-specific variability, demand fluctuation, inflation, and cost-impacting ESG (environmental, social, and corporate governance) initiatives are a few of these. The Industry 4.0 promise to deliver significant cost and revenue efficiencies leads decision makers to undertake strategic digital transformation initiatives. For example, a hypothetical US consumer electronics company with a global production network could offset rising carbon emission offset costs and taxes with aggressive automation, according to a recent McKinsey analysis.

However, according to another McKinsey study most digital transformation initiatives fail to achieve enterprise scale. If successful they would deliver huge economic benefits with 10-30% increased throughput, 15-30% increased labor productivity, 30-50% reduction in machine downtime and a whopping 85% more accurate forecasting.

To increase the chances of digital transformation success at factory networks and manufacturing processes at scale, it’s essential to have:

• Composable business goals: Defined key business goals – reduced product lifecycle times, on-demand manufacturing, reduced waste, reduced spend per factory, etc. – early in the process and refined along the way.
• Composable architecture-driven implementation: Iterative, agile, and smart manufacturing methods responsive to ecosystem changes combined with centralized governance.

To achieve smart, modular, agile manufacturing at scale and with ease, speed, and reduced cost requires digital technologies that allow customers to compose, decompose, or re-compose manufacturing and factory functions and snap them together, akin to arranging the blocks.

The Limitations of Existing MES Architectures

Consider the following scenarios:

• An automotive parts supplier has recently reduced its five-step gear-box manufacturing process to three steps, reducing operator stations.
• A large agrochemical manufacturer recently upgraded its ERP system, resulting in changes to the data-transfer payload formats and content.
• A large semiconductor company chose to include an AI-vision-based quality inspection layer in a closed loop manufacturing process, causing the new sensors to save more variables and data points into the historian.
• A biosimilar manufacturer must modify the “safety efficacy” test protocols as per FDA’s new guidelines, requiring an enhancement to the Reporting module.
• A food processing company must include in-process material verification to achieve greater control over material tracking and waste management.
• The factories in a specific geography of a paper and pulp company chose to replace a group of assets, resulting in the manufacturing execution system interacting with a new class of PLCs (programmable logic controllers) on new assets.

What do the above requirements have in common?

• These are significant, but common requirements for which:
  • Companies incur heavy IT costs to update the manufacturing execution and operation systems to accomplish them
  • Long IT implementation cycles cause downtimes and product manufacturing delays. On average, it takes six to 8 months to implement a “significant requirement,” as mentioned above, followed testing and deployment.

On average, 2-5 full-time resources are required to spend 6 to 8 months to implement one such requirement, followed an additional time for testing and production deployment.

Such cost and time impacts are majorly induced the constraints of the software products built on traditional architectures.

Monolith – A rigid architecture where the layers of the product and the inner workings of functional components are intertwined. Making changes to one function exposes the rest of the system to potential regression, mandating significant downtime for testing and validation a large contingent of resources with varied skills.

Modular / Microservice – A modular approach allows deployment and management of capabilities/modules on demand. However, if the requirements diverge from the designed interface boundaries of modules, customers will be forced to build a custom “spaghetti” integration layer with duplicated meta-data driven canonical interface models. Additionally, the module boundaries are defined at design time. So, the post-built custom extensions impart high cost.

Custom – Purpose-built solutions are not built with broader (industry, domain, or technology) best practices but simply replicate a customer’s existing business process. The highly probable project delays and increasing implementation costs create uncertainty and risk to strategic goals. An additional cost of regular enhancements drags down the total cost of ownership (TCO).

Most, if not all, traditional manufacturing execution systems employ one of the above architectures and thus struggle to help customers realize the value of digital transformation for manufacturing at scale, speed, with cost efficiency.

However, a composable “blocks” architecture could reduce the cost and time to achieve the “significant” changes to manufacturing systems. This in turn, enables composable, agile, and smart manufacturing goals.

A Modular, Composable Approach

A composable solution is made up of building blocks of functions that are carefully designed with the following principles:

1. The form, fit and function of each building block are small enough to be easily decoupled from a larger assembly. However, each building block must be large enough to complete a significant “atomic” activity (e.g., a function such as kitting, grinding, or mixing, or a PLC connector for a specific version of a particular vendor with the ability to query certain “configurable”—i.e. not pre-defined—tags via configurable interfaces).
2. A configurable interface (input and output) layer allows replacing one “block” with the other. The interfaces should be query and event-based, easily configurable post-deployment, and without ‘codeful’ changes.
3. In-memory storage with an integration layer can handle in-flight data efficiently consuming compute and storage resources – e.g., data payload size or fault-handling standards. Offers a best-in-class integration model.
4. Separation of resources -separate memory, CPU cycles, and other resource footprints among the ‘blocks.”
5. Support for low-latency and high-latency capabilities – On a manufacturing shop floor, connecting with L1 and L2 systems such as IoT devices, PLCs, or DCS is critical at low latency and in a synchronous manner, whereas connecting with L4 systems requires high latency and an asynchronous communication capability.
6. The overall product platform architecture requires the presentation, business logic, and data persistence layers should be loosely coupled so any layer can be replaced with minimal impact on other layers. Any change impact is minimal and can be customized per customer needs without turning into a bespoke solution.
7. A model-based, simple, drag-and-drop workflow to build the business logic that reduces or eliminates the need for software developers, instead enabling the domain experts to define the business logic with ease and intuitiveness.

A platform with such characteristics enables composable manufacturing. This is a game changer in addressing most other architectures’ failures in enabling digital transformation at scale, speed, and reduced cost.

Key Benefits

• Reduce the impact of change: The first-time assembly of key  blocks and ensuing composition, decomposition, and re-composition of  blocks takes days to weeks instead of months.
• Standardize and scale, but enable plant-specific customization: Scaling across a factory network with a centralized, certified assembly, with an ability to apply factory-specific  blocks as deployed to factories, is designed as an automated process via a centralized enterprise management model.
• Reduce validation time: Customers requiring heavy quality, regression, and compliance testing cycles for any significant product configuration changes will significantly reduce effort as the blocks can be disjoined business logic and data manipulation.
• Reduce downtime to take change: No required downtime for replacing, adding, or removing the blocks in some instances, resulting in high efficiency.
• Organic scaling across factory networks, versus imposed standards: Spreading the digital transformation becomes more organic as new blocks can be quickly created to achieve “simple single function” factory operations – such as bar-code scanning, truck dispatch tracking, etc.

Conclusion

Smart, agile manufacturing that drives dynamic decision making and responsiveness to ecosystem challenges requires a composable solution. In other words, composable manufacturing will drive the scale and speed of digital transformation for manufacturers with reduced costs.

The manufacturers who adopt composable manufacturing technology will invariably:

1. Reduce the impact of change.
2. Reduce the cost of design, implementation, and deployment.
3. Reduce time to value for imparting change.
4. Achieve digital transformation at scale and speed.