“Supply chain agility can be defined as an organization’s ability to profitably manufacture and deliver a broad range of high-quality products and services with short lead times and varying volumes while providing enhanced value to customers.” – The Hacket Group “understanding the benefits of supply chain agility”
Agility is directly linked to the ability of equipment, processes, and systems to deal with variability. The historical approach to eliminating variability has served us well in the past.
Eliminate variability so we can be in our happy steady state of operations! Building a truly agile supply chain requires being prepared for the fact that a steady state will be constantly challenged; it requires embracing variability.
Successful implementation of agility is about achieving new steady states in response to market needs and exploiting opportunities offered by the raw material supply – quickly, and without any loss of optimal efficiency.
Ideally, it should be more than being able to respond to market needs. It should allow the business to aggressively lead the creation of new market opportunities such that less agile competitors are placed at a disadvantage.
Achieving this state of agility is a business transformation initiative. This requires preparing equipment, processes, and systems which can adjust to, and withstand variability. Another way of looking at it is the responsiveness of manufacturing to the demands of agility.
Digitalization of manufacturing is the key enabler for this preparedness. When implemented with this approach – the digital transformation (DX) of the business – ROI can be clearly linked to enterprise objectives.
Building a data and information system for agile, smart manufacturing requires two key elements – visibility and controllability.
The gap between the C-Suite and the plant floor is one of the biggest challenges to building a data-driven supply chain. Continuous improvement teams at the plant level are working on specific challenges like uptime and the executive team is seeking a large-scale digital transformation to create new business models.
There is a gap between what these two groups are doing, and they are often not viewing the same issues. This should not be viewed as a cultural issue (although that can be the case sometimes). It is a visibility issue. Given the nature of their roles, they have to have different perspectives.
This gap directly impacts agility. The ability to overcome this gap is a function of:
- How well the plant-level perspective is represented in the Management Operating Systems (MoS).
- How capable the plants are to generate the information needed to adequately inform the MoS.
The ideal state is a real-time, or near real-time visibility and connectivity between the two. The state of existing automation and data systems are generally the big barriers.
A high degree of controllability is required to achieve higher agility. Controllability is impacted by:
- How responsive the plants are to the MoS?
- The degree of automation of the plants
- The flexibility of the manufacturing processes
The ideal state is to add process control system-like functionality to the MoS – giving the MoS levers that are directly connected to the plants. Since the role of the MoS is to serve the executive teams, we refer to this control system functions as a Value Control System (VCS).
We’ll describe how such as VCS can be added to the MoS.
Requirements, barriers, and opportunities
Both visibility and controllability between the MoS and the plants require creating new ways or adapting the existing systems to communicate in a different language. This language requires higher abstractions created from data with analytics – both streaming, as well as batch analytics.
The manufacturing systems on the plant floor including plant automation and MES systems need to be able to communicate the product and equipment state, in real-time and predicted from OT & IIoT systems, in the form of key metrics abstractions that the MoS can easily consume for decision support.
These systems also need to be able to consume the commands from the MoS-VCS and translate them to lower-level OT and automation systems.
The creation of this language of analytics requires the creation of context and presenting it to different perspectives needed in the MoS.
Context – Everything is temporal!
The general nature of manufacturing data is time-series data. This is well understood from the plant floor equipment, sensors, and “IoT” perspective – in terms of the equipment used to make the product; but largely misunderstood from the procedural and process perspective – how the product is made.
The procedural perspective, generally the purview of ERP and MES systems is largely seen as transactional. Therein lies the big challenge, as well as the big opportunity to create a digitally aware manufacturing enterprise.
Creating a temporal context between the equipment data, procedural, and transactional data is the foundation of a digitalized manufacturing enterprise. The key challenges to creating this context are:
- Highly heterogeneous data:
Given the very purpose of the systems at the higher levels of the enterprise, such as ERP systems, the time horizon of the data is long – such as product campaigns or customer-specific transactions that may span months; or MES systems managing shift or daily production schedules.
The time horizon of the plant floor systems on the other hand may have time horizon spans of minutes, seconds, or even sub-seconds.
Different shared manufacturing equipment at the same plant may be generating data related to different transactions requested by the ERP and MES systems at the same time.
- Siloed data
The data generated by different systems concerned with different aspects of the operation generally reside in different systems – such as MoS, ERP and MES systems (IT), and Control/SCADA systems (OT). On the plant floor itself, data of the same time domain may also reside in different silos.
Opportunities on the plant floor:
- Responsiveness to variability, introduced by
- Raw materials – real-time understanding of incoming material variability
- Manual steps between unit operations
- Flexibility in asset utilization/reconfiguration
- Speed/setup time of changeovers
- In-process inspections vs manual inspection
- Batch->Contiguous->continuous operations
For a more in-depth look at the challenges of industrial data in the context of machine learning and streaming analytics, here is a post from my colleague, Pranav Prakash (@pranav9278): https://www.quartic.ai/blog/the-nature-of-industrial-data/
Common data-related challenges on the plant floor
- Poor or non-existent connectivity between ERP-MES systems and the plant automation systems.
- Islands of automation
- Not all data generated for the purpose of controlling the equipment is collected in plant data historians.
- Many facilities are built using OEM-supplied equipment typically provided with automation sufficient to adequately control a piece of equipment or at best a unit operation. While it is common to connect these islands of automation for some plant-level visibility with a SCADA system, generally the amount of data connected may not be sufficient to create the analytics-based visibility required for agility.
In the subsequent section of this series, we’ll address how to overcome these challenges, and embrace variability to build a highly agile manufacturing supply chain.