I got an email from an old customer that I had done a Finite Capacity Scheduling (FCS) implementation at a few years back, inquiring about whether I knew if there had been any enhancements made to the product with regards to improving its ability to take material supply issues into account when scheduling.
After responding I no longer implemented Finite Capacity Scheduling systems as I had found a superior solution for production planning (and was tired of unmet expectations on the customers side due to the limitations of FCS), I started thinking about the best approach to the finite capacity problem encountered in many manufacturing environments.
The problem of bottlenecks in production is one that is common to all industries in all countries, no matter what their business. The question of the best way to deal with this problem has long been an issue for ERP system vendors and systems consultants alike. In order to address this critical issue in production planning, we need to examine the factors that go into making a planning tool effective for the user when dealing with finite or constrained capacity.
The main features required for an effective tool are as follows:
- The ability to model multiple types of constraints in the production process in a clearly understood manner
- The ability to integrate material and production capacity constraints into a single planning model
- The ability to collaborate with other users in the production process to ensure the best plan possible
- The ability to easily integrate and centralize data from multiple sources in a single system
- The ability to quickly and easily change demand and supply scenario’s in order to determine the optimal plan
- The ability to react quickly and effectively to constantly changing real world events
If we use these requirements as a starting point, it quickly becomes apparent that many Finite Scheduling tools fall down in several areas. First of all, most FCS systems are designed from the production scheduling viewpoint, leaving materials and outsourced production constraints as a secondary consideration, if considered at all. Since current manufacturers rely heavily on outsourced procurement and purchased materials, this lack of deep functionality tips the scales to constrained planning systems. Further to this point, a constrained planning tool which also has sophisticated Demand Planning capabilities integrated into the tool moves the argument even further in favour of Constraint planning. FCS tools tend to narrowly focus on detailed capacity constraints (changeover sequence, etc), which require a great deal of complex modelling and setup to properly emulate real world scenario’s. Most of these bottleneck issues can be satisfactorily addressed with simple constraint modelling if viewed from the perspective of the entire supply chain, thereby greatly reducing implementation complexity and allowing a holistic approach to supply chain management.
The next thing we need to look at is scalability and the ability for multiple users to simultaneously access the tool from multiple parts of the globe. This requires a data engine which can handle the complex calculations required in finite or constrained scheduling along with the ability to accommodate the large amount of data which is required in order to give a truly accurate picture. Because of the complexity of constrained systems, multiple users in the supply chain must be able to work together in a collaborative manner in order to move the planning process forward. The most accessible and cost effective answer to this requirement is to use the Internet as a platform on which to construct a broad based, secure solution. In other words, the tool must support current state of the art security and data transmission technologies over the Web, and allow users to access the tool via a web based interface.
Many vendors tout the benefits of a single platform, which from the transactional side of business cannot be questioned. However, finite capacity and constraint planning are not just transactional functions, but require advanced functionality only found in systems designed specifically for planning. Since most FCS systems are very much geared to the production process in terms of integration, the nod must once again go to full featured constraint planning systems with its wider integration capabilities.
Both full featured constrained planning systems and FCS systems have visual simulation and user friendly what if capabilities. However, the ability to compare different scenarios is critical for the development of an optimized plan, and FCS systems with their Gantt chart displays have difficulties meeting this requirement. A constrained planning system with robust score carding capability comes out ahead in this regard.
Lastly, both types of tools support rapid changes to the plan thru sophisticated user interfaces, but only the constrained planning system incorporates the demand and materials planning capability required to reflect all changes in the supply chain effectively.
In summary, when all relevant factors are considered in the selection and implementation of a scheduling system, it can be clearly demonstrated that the full featured constrained planning system will give you a better tool to manage the entire supply chain than will a production focused FCS tool, resulting in a better ROI for the project. As well, most production bottleneck issues can be solved with simple constraint based models, so why limit yourself to addressing one segment of your supply chain when a holistic approach can bring much greater benefits to the organization as a whole.
TrackBack
• Digg This
• Add to del.icio.us
Tags: Advanced planning & scheduling (APS), Capacity
Posted in Best practices, Supply chain management
You can leave a response, or trackback from your own site.
I certainly am not going to disagree with your main conclusion. When we look at the business problem, it is essential a linear (or non-linear) programming problem. The difficulty has always been that there are so many variables and it is so difficult to set the coefficients for the variables, that solving this problem is almost impossible.
Without getting too geeky, much of the work of mathematicians in the latter half of the 19th century and early 20th was to show that the world (as defined by mathematics) was not deterministic. That is, given a set of initial conditions, I can’t necessarily extrapolate any distance into the future. And that is exactly what we are trying to do with our capacity systems (and with MRP II in general).
The truth seems to be that the horizon that I can control is very short. And we have labored mightly over the past decade to make that horizon even shorter. And many of the decisions we have made (single source vendors, etc.) have made it harder to respond within that short time.
Since I can only control a short horizon I think the emphasis today has to be on reducing the manufacturing horizon (the lead time), and providing some flexibility in case things go all to pieces (if we can’t get something from one guy we need to get it from another or else stockpile it ourselves inventory turns be damned). I really feel that strategy has to be an integral part of any kind of capacity system. The last ten years have proven what Godel knew – you can’t predict the future for very long.
Hi Martin,
+15 years experience of implementing APS systems have made me draw exactly the same conclusions as you. Often consultants and clients don’t see the forest for the trees…………
rgds
Lars
The modeling of a very simple business flow can be a very taxing and complicated process. Typically systems allow only a single quantitative figure to be input to represent the capacity of any given activity. If one was to more acurately describe the capacity it would consist of more complete statistical representation of average capacity, deviation around the average, and shape of the deviation. Computer models which seek to utilize these broader factors are available but the users must be sophisticated in order to make use of them.
I very much agree with Dave. The real success factor lies in reducing the period of uncertaintity to the degree that it economically feasible. Each business must analyze the risk/reward related to this process.
Software is not the answer to all problems. In businesses which have a significant degree of uncertaintity involved in their supply chain, it may be that surplus capacity and/or inventory is a more realistic approach to responding to variation than investing in computer modeling which attempt to forecast the variation.
I’m of two minds on Finite Capacity Scheduling and Constraint Based Planning. If a factory needs to implement these methods, it may mean the factory layout and flow is too complex for a person to understand–implying the constraint migrates to all parts of the production processes. F
For people who run factories it is important for them to identify what is the most valuable and expensive resource in the plant so they can make plans to protect it when things go wrong. FCS and CBP tend to become black boxes that do their calculation and present solutions that will be close to optimal but will confuse the operators. I would rather see factories simplify their production flow so it is very clear what the critical path is, and then consider implementing FCS or CBP. First, the factory will be more understandable to the operators and, second, it will make the solutions much easier for the analytical engines.
So I am all for optimizing factories and I tend to lean toward constraint based methods, but effort should be spent on simplifying the flow and ensuring the constraint is the constraint. In other words, as Goldratt would say subordinate all other operations to the necessity to exploit the constraint.
Ben
It concerns me when I read so much information regarding the planning process. So many
systems are implemented where it is taken for granted they work as intended. In recent times
I have asked whether there are the proper inspections to see if the intended effects are actual.
Although I have to agree that FCS systems fall short, this article ignores an intermediate solution: Advanced Scheduling systems.
Within APS (Advanced Planning and Scheduling) the better Constrained Planning (CP) systems fall under the Advanced Planning umbrella. Solutions of this type typically have a longer time horizon and higher level view of the underlying data as indicated in the article with all the described benefits.
Advanced Scheduling systems fill the gap where a broader coverage of the supply chain is needed (yet still within the confines of the own company), but a high level of detail is required. These systems cover all of FCS footprint but extend over material call-offs including all the constraints on these.
For example I encountered a scenario where bulk product was loaded in compartmentalized train wagons. Due to contamination constraints, the train compartments, the pipes feeding each compartment and the silos from which the pipes fed product all were under very tight scheduling constraints that require a solution of high level of detail. FCS does not cover the pipes and the train compartments, CP can cover them but lacks the level of detail needed, yet Advanced Scheduling solutions exist that cover the requirements perfectly.
Admittedly, this is a rare scenario, but there are other common scenarios that would benefit equally from breadth and detail of Advanced Scheduling.
I appreciate all the comments posted, and I do not see anything I can argue with. My main point at the base of it all is that a holistic approach to planning in the supply chain is preferable to a piecemeal approach to try and focus one’s energy on sections of it. When dealing with limited resources, the best ROI is achieved by integrating your supply chain, so all pieces work together in the most efficient manner possible. Part of this process should involve simplification and optimization of each piece so they can be integrated with minimum effort and into a single entity. In order to manage this holistic supply chain, a user friendly software solution that can model the complete chain is mandatory.
I think you might find that a group has solved the capacity and material constraints dilemma. Constraints Management Group proposes that Actively Synchronized Replenishment (deals with material constraints) can be combined with a Dum-Buffer-Rope model. Very interesting paper can be found at http://www.thoughtwarepeople.com.
I’m glad you clarified you point in #7. With all the fancy “Buzz” words flying around I totally missed that. Am I missing something, are there folks out there that are not selecting, deploying or upgrading end to end business systems for their supply chain?
I’ve been doing this for an awfully long time and it just doesn’t seem that complicated to me. Dead nuts your planning factors in conjunction with how aggressive you want to get with your inventory turns, “manage” your supplier relationships to where you can bank on what they tell you mixed in with a little good old fashioned management by walking around….
Oh yea, and have your staff read “The Goal” – it’s a little less confusing.
Hi Martin,
Working for a company making End to End Constrained planning tool I agree with you on the superior ROI, value etc. However there are situations that require you to choose a Finite Capacity Scheduling tool.
The Limitation of an END to END linear programming based tool on a complex supply chain network is run time especially when you model factory flows with complex relationships between them. In discrete parts assembly systems like an auto assembly the work centers are capacity constrained and also constrained by the sequence of the flow required. The run time grows dramatically with the complexity.
In manufacturing set ups where the process complexity is high, and variability of both supplies and demand is high and reworking of planning is very often an Advanced Planning Tool may be unsuitable. Here Daily Scheduling may be better off with a quick hueristic based Finite Scheduler
Thanks for your input, Samit. I do agree with you that in a minority of cases, a Finite Scheduler is the best solution, especially when you are concentrating on solving a complex scheduling issue in a high production speed, high changeover environment (reducing downtime, increasing machine utilization, etc.).
However, the run time issue can be overcome with a memory resident tool, utlisuing a fully in-memory database to significantly speed up the calculations required . As a consultant for Kinaxis, which produces a tool called RapidResponse, I know these tools can run complex analytics on large amounts of data in a matter of seconds. This ability allows the user to look at multiple what if scenario’s in a small amount of time, and gives them the ability to optimize planning across the entire supply and demand chain, not just the production floor. This also allows for a quick response to changing conditions in the business environment.
Hi,
helping a client right now to implement a solution from http://www.optimity.com.au
It is constraint based planning using optimization but what strikes me is the extreme ease of use. No complex and hard to understand parameters but very straight forward configuration once you have defined ABC classes for products and customers. Ideal for companies who do need to consider supply constraints as part of the replenishment planning process.