Part 5: Holistic Production Planning is Essential

End-To-End Modeling & Digital Twins

Incorporating the end-to-end production process into the vision for a supply chain planning model is vital to attaining the scenario modeling flexibility and profit optimization that many supply chain stakeholders envision. Upon considering modeling the entire operation, many readers may envision a digital twin. Technically, by the time an entire operation is holistically and accurately modeled, what you’ll have is essentially a digital twin—but building a digital twin doesn’t have to be an end goal. Even if it is, a manufacturer doesn’t have to master a digital twin or perfectly model their entire production operation on day one or at the time of detailed scheduling or S&OP/IBP implementation. However, choosing a technology that has the capability to do so is essential as it establishes the platform that enables the supply chain team and stakeholders to iteratively improve the models, inputs, and outputs as they inch toward perfection. Likewise, supply chain excellence is almost never achieved with one project, initiative, or technology implementation—it typically occurs over years and in a supportive culture with skilled people given the right resources (including appropriate technology) that enable iterative improvement.

The reality is accurate modeling takes time because the business will change as it becomes more efficient. Establishing a highly accurate and holistic model occurs in more of an agile method rather than a big bang approach or a one-and-done project. This is also how higher levels of supply chain and operations excellence are typically achieved—methodically and over time. The key is that once a well-fitting system is implemented, it sets the stage for the organization to progress to those higher levels of modeling accuracy and performance at its own pace and with no further technology changes required .Once the right platform is in place, the agile refinements that hat make the model increasingly accurate and holistic typically involve little more than replacing a data set or making a few user-configured adjustments.

In a practical example, the most critical and perhaps most basic portions of the operation may be the only ones accounted for initially. In that case, the S&OP process and scenario modeling will work with the fidelity and flexibility that model “Version 1.0” provides. Over time, as improvements in supply chain modeling and operations occur, augmenting the model with parts of the production operation previously perceived as too complex, begin to appear attainable and the benefits of doing so become more readily apparent. Thus by simply configuring and tweaking the already implemented system/platform, users and stakeholders can drive the model from Version 1.0 to Version 2.0 to perhaps a Version 10.0 and beyond as each enhancement is mastered and the next high-value incremental improvement comes into focus.

Conversely, if the technology cannot handle the holistic operation before hitting its own internal design limitations, these incremental improvements will often never take shape as there would be no way to test, observe, implement, and build upon enhancements that users may begin to envision as supply chain accuracy improves. In this case, the movement toward increasingly higher levels of operational and supply chain excellence is slowed, halted, or perhaps even never takes off as users hit an artificially low ceiling in terms of what they can achieve with existing resources and technology.

In many cases vendors that have technologies which cannot address an entire operation will recommend only addressing the portion of the operation they can support and may suggest that modeling the rest of the operation is unnecessary or unimportant from a scheduling or scenario planning standpoint. Since most vendors tend hit their technical complexity limit around the upper end of the discrete manufacturing range, this logic is commonly used to limit the scope to the discrete tail of a process manufacturing operation (i.e. starting with the switchover point where production transitions from continuous to discrete) or the last few steps of a complex discrete manufacturing operation.

This is illustrated in Diagram 5.1 which depicts sample operations for distributors/resellers, discrete manufacturers, and process manufacturers. For each type of operation, the orange diamond represents “Planning Point 0” where a given vendor (and sometimes the business stakeholders themselves) might suggest the planning scope should begin.

For distributors and resellers, the challenge is not so much a production modeling issue as it is an inbound-outbound and inventory balancing problem. There are unique challenges and complexities that distributors and reseller supply chains have any that any planning solution must address, but these challenges are of a very different, less technical nature than modeling complex production environments. Likewise, for resellers and distributors, most S&OP/IBP technology vendors can support plans spanning the entire range including inbound transport (product vendor inventory and replenishment), storage, and outbound transport to customers or a specific retail location.

Diagram 5.1 – Typical Flows by Industry Group vs. Start of Detailed Scheduling/S&OP/IBP Planning

For discrete manufacturers, each step or workstation in the production process has material and labor inputs. Some vendors may recommend limiting the scope to the last several steps or assembly points in the operation if extending upstream beyond that point exceeds the capability of the vendor’s solution. In this situation, the modeling and planning scope also starts at Planning Point 0, accounting only for the last few steps in the operation. This last portion could encompass one, two, or twenty steps, but the key is that it doesn’t account for the entire operation and is therefore not a holistic solution.

Every process manufacturer is, to some extent, a hybrid operation as the continuous portion of production is eventually converted to discrete units (e.g. bottles of soda, barrels of chemicals, reams of paper, bars of chocolate, etc.). Vendors whose solutions cannot accommodate process industry constraints will often recommend beginning the modeling and planning scope at the start of the discrete portion of the manufacturing process—also represented as Planning Point 0. In process production environments, sometimes the decision-makers themselves initially view the upstream process as being too complex to model holistically. In either case, it is a choice that places a hard limit on potential supply chain improvements and undermines serious advancement toward Industry 4.0 capabilities. Most importantly, this partial approach will eventually prove problematic in addressing even regular day-to-day operations. Consider the number of times that problems or dependencies in the upstream (or process) portion of a plant created production problems that required adjusting the entire production plan including downstream (or discrete) portions of production. While the process portion of a manufacturing operation can be somewhat more challenging to model, as with any journey, it begins with a single step and bit-by-bit, what seemed insurmountable is eventually tamed. At this point, the organization reaps tremendous operational and financial benefits from truly holistic and flexible supply chain planning and optimization.

Diagram 5.2 – Illustration of Iterative Model Development

Diagram 5.2 illustrates this iterative approach for a hypothetical process manufacturer with a seven-stage production process—the last three stages of which are discrete. In each stage, the model might temporarily use a pass-thru, effectively ignoring the stage, or the level of detail can improve progressively from high-level to an exact representation. Since it isn’t abnormal for more complex stages to lag other simpler stages in modeling, Version 1 of this model only covers stages 5, 6, and 7 at a moderate or exact level of detail with other stages being pass-thru or only represented at a high level. By Version 2, the models for production stages 1, 4, and 5 have each progressed a level with additional progression made in Version 3. Finally, by Version 4, the model for every stage is at the most accurate “exact” level of detail, except for production stage 1 which remains at a moderate level. This iterative progression is commonly how model development happens, but such progression would be rendered impossible if the technology is unable to support many of the production stages.

The Devil in the Details: Two Real-World Examples

The limitations of a partial model may seem subtle or even irrelevant, but the impacts can be significant or even considered severe. To help illustrate this concept, it’s worthwhile to explore two real-world examples.

Example #1 – A single constraint, even with the “Planning Point 0” scope limited to the discrete portion of a process production operation, effectively broke the vendor’s scheduling solution.

Situation: Consider the case of the client “case study” I provided in the introduction of this article. In that case, the client was a process manufacturer and a critical disconnect was that their selected supply chain planning technology only understood independent production lines. However, the client’s production facility was configured such that a single pipe from the final processing tank fed two lines on the production floor where molten composite resin was extruded, cooled, and cut into discrete pieces of material. Because the two extrusion lines were tied to the same feed pipe, each line could only produce SKUs that required the same formula of molten resin at any point in time. Beyond that constraint, each line could independently create any of multiple SKUs associated with that one molten resin batch including different sizes, shapes, etc.

Limitation and Attempted Fix: The supply chain planning vendor’s scheduling solution was not designed to account for this type of constraint or most other process-oriented constraints, for that matter. This was also not a configuration issue as the limitations were baked into the core design of the software. Upon implementation, the only way to have the vendor’s system create a realistic schedule involved guiding the vendor through a re-write of their core software so that it would factor in that constraint when it created the production schedule. Problem solved, right? Nope. By introducing that “patch” into the vendor’s software it rendered a lot of the other functionality inoperable but there was no practical way around that. The functionality that became inoperable included automatic adjustments to previously created schedules (e.g. maintenance time, R&D runs, etc.), scenario modeling, and intra- and inter-plant optimization among others. And because these limitations were at the very heart of the software’s design, to correct the system in a way that accounted for this seemingly simple constraint while maintaining all of the application’s functionality would have required a near complete re-write of the vendor’s scheduling and S&OP application. We settled with the limited patch since doing a nearly complete re-build of the vendor’s application would introduce a number of other problems.

Outcome: The plant scheduler could create the initial schedule using the planning system, but any of the frequent subsequent adjustments had to be done manually. Thus, the time benefit to the scheduler was potentially negative as they now had to do the work in the planning system as well as using the original manual approach. The benefit expected by R&D and maintenance was lost as their planning and input still had to occur manually with all the shortcomings that involved. Finally, the operational and financial benefit of scenario modeling, cross-plant planning, and optimization was substantially undermined, even though a sizeable benefit was still realized due to changes in the scheduling logic which the vendor’s application supported. This one benefit was the only reason it still made sense to create the initial schedule within the application although a simpler, cheaper custom application could have performed this single task. The client wisely chose to use a third-party supply chain planning system largely for the broad benefits they offer (i.e. scenario-modeling, schedule updates, proactive maintenance and R&D planning, multi-plant planning, flexibility, optimization, etc.). Yet, because they selected a technology that could not accommodate the type of constraints that existed in their process manufacturing environment, they lost most of those additional benefits.

Political Considerations: There was still enough benefit attained from the S&OP process and the new scheduling methodology that the project could have been viewed as a success. It’s also possible that senior leadership was not fully aware of the significant problem with the selected technology or the foregone opportunities that entailed. It would have been most comfortable for the sponsor and key stakeholders to focus on the achievements and minimize the problems and missed ancillary benefits. From a CxO or executive leadership perspective, evidence of any problem could be limited to grousing from plants about having to use the cumbersome new technology since two of the key benefits were still attained. The point here is that the attainment of some partial operational and financial benefits can mask the potentially large opportunity cost of making a bad technology decision.

Example #2 – A major vendor’s solution with scope limited to the discrete portion of a process production operation could not accommodate bi-directional planning in an environment where “downstream” logistics constraints would need to inform “upstream” production schedules.

Situation: A process manufacturing company was implementing the detailed scheduling layer solution at multiple plants, starting with Plant #1 and proceeding to Plant #2, Plant #3, and so on. The selected vendor’s solution was not capable of handling the higher technical complexity associated with process industry constraints, and both the vendor and manufacturer decided it was only necessary to account for the discrete portion of the production process.

Limitation: Besides only being capable of addressing discrete production operations with, at best, moderate technical complexity, this vendor’s scheduling solution was also only capable of forward-only/linear planning. Specifically, it would use changeover cost/priority tables to establish the best schedule at one production point, then move to the next point and apply the same logic there, and so on. See Part 2 / Principle 2 of this article for a more in-depth explanation of the problems with this planning approach. While Plants #1 and #2 did not have significant downstream constraints that would impact upstream schedules, Plant #3 plant did have downstream capacity and logistics constraints that required adjusting upstream production schedules. However, the forward-only/linear planning approach used by the system would render this impossible without a “patch” or some form of customization.

Impacts: Upon recognizing the system was incapable of reverse planning, the supply chain leader at Plant #3 confirmed that the detailed scheduling system would often produce an unworkable schedule so they would need to continue planning manually to accommodate the constraints at that plant. Also, when one or more plants (in this case Plant #3) is not able to rely on the supply chain planning solution because the solution cannot accommodate key constraints at a plant, then the corporate level multi-plant planning, scenario modeling, and optimization benefits of the supply chain planning solution can also become severely undermined.

It’s worthwhile to note that the supply chain planning vendors used in each of these examples have both consistently appeared in the “Leader” quadrant of Gartner’s supply chain planning-related Magic Quadrants. And based on marketing materials and sales pitches, these two vendors’ capabilities can appear nearly indistinguishable from the capabilities of vendors whose solutions could gracefully accommodate both situations, thus delivering substantially greater operational and financial benefits.

Longer-Term Implications

Companies do not achieve supply chain excellence with one project or technology implementation, it comes with incremental improvements at various points in the process over time. Each advancement nudges the supply chain performance level and financial benefits to an increasingly higher plane until the entire supply chain function is a leading-edge operation delivering outsized value. Likewise, supply chain planning, detailed scheduling, or S&OP/IBP solutions do not necessarily have to address the entire process from day one, but they should have the capability to do so. If they cannot address the entire process, the creative sparks that would otherwise lead to incremental improvements in completeness and accuracy never take hold, and neither do the additional operational and financial benefits.

Also, as discussed in the prior section, most Industry 4.0 features and benefits for manufacturers are based on holistically modeling the production operation. This doesn’t mean that every single one of the most minute details of a production process has to be systematically and painstakingly replicated. Many of the more detailed processes can be kept at a high level and made more granular only as the organization is ready and deems that it makes sense to do so. But having that holistic model does require an end-to-end programmatic understanding of the operation without massive gaps, even if the more complex portions are generalized at first and refined later.

Hypothetically, a manufacturer could implement an ill-fitting or atomistic detailed scheduling or S&OP/IBP solution with a plan to replace it later when they feel they have outgrown it, but that’s usually not a realistic or strategically astute view. The more likely reality is that the ill-fitted solution is going to remain in place for several years as the time, cost, learning curve, and politics will often prevent organizations from making a change so soon after implementation. During that time, it can insidiously restrict the organization’s growth to the point that they may never actually recognize that they’ve outgrown it.

Alternatively, the limitations could become so great that it needs to be replaced out of operational necessity or the team will resort to manually working around it. Neither is a desirable outcome. And when an ill-fitting or atomistic system is replaced, there is still the organizational distraction and expense associated with even a relatively simple supply chain planning system implementation. Why plan for operational limitations, stunted supply chain development, restrained benefits, and future expense if you can avoid them?

There is currently little to no difference in the license or implementation cost between supply chain solutions that are well-suited to accurately and holistically address a given manufacturing environment and those that have significant modeling, planning, and optimization limitations. Perhaps this is because there is so much misinformation and the benefits and risks are often not recognized or acted upon. Whatever the cause, this another reason why it makes sense to choose a vendor solution that can accurately and holistically accommodate the manufacturing environment from day one—providing the platform for boundless supply chain advancement and the operational and financial benefits that go along with it.

Do you have any thoughts, additional perspectives, or different views? Let me know what you think in the comments.

This due its length, this article is published online in multiple parts. Click on the sections below to view other parts of this article, or download the full PDF document immediately by requesting the download link in the box below.

> Next Section: Part 6 - Accurate Modeling is Top Priority

< Previous Section: Part 4 - The Subtle Blast of a Solver’s Approach

Table of Contents