"My supply chain is different and more complex"...if I had a dollar for every time that I heard this from potential clients, I would be a millionaire by now! Yes, your supply chain is unique and different but it does not mean that it is more complex than others. Every industry has its own nuances of how the supply chain can impact their business. Unless you are dealing with niche players, planning systems are designed to handle the complexity of all supply chains. However, what makes them different is the level of accuracy that they have in order to model the supply chain. With a few exceptions, all S&OP solutions treat the supply planning just like a spreadsheet does. The use bucketed capacities, static lead-times, and predefined bottleneck resources! At the level of spreadsheet modeling obviously they can handle any supply chain just like spreadsheets have been doing for a long time. If, however you are interested in a reliable plan for the supply chain then S&OP by itself is not good enough; you need S&OE to enable you to execute the plan. What good is a plan that needs to be manually changed over and over again in order to make it work? In the words of Gartner: “The No. 1 challenge among manufacturing companies is connecting sales and operations planning (S&OP) to operational plan/execution.” To learn more about how to generate reliable plans that can be executed click here: S&OE.
Like many other S&OP solutions, SAP®-IBP* is designed to perform high level “rough cut” planning. On the supply side, it uses a simplistic logic that has been used in spreadsheets, for decades, to plan production. It assumes fixed leadtimes, pre-defined bottleneck resources and bucketed capacities. This is not much different than the old MRP methodology from the 70’s, but now categorized under S&OP. Advanced planning came to play because this was not effective and plans were not accurate enough to model the capacities and coordination of material and capacity was missing. SAP-IBP, much like almost all other current S&OP solutions, is intended for very high level reporting. Once the plan is generated it can be at most 60% accurate. Hence, once given to the operations side of the business, it does not work. SAP’s answer to this problem is usually faster what-if analysis, which still adds up to lots of manual adjustments, much like working with spreadsheets to make unworkable plans to work.
This is exactly what the APS industry felt the need to change with the use of powerful optimization engines in the late 90’s followed by AI techniques in the past 15 years. Yes, SAP-IBP provides nice reports and charts, and maybe even help in collaboration, but in the final analysis, it lacks realistic plans and cannot feasibly translate generated plans into actual execution of those plans. In other words, it misses the mark when it comes to Sales and Operation Execution (S&OE). Specifically, it lacks proper modeling of resources to understand true capacities, it fails to estimate true leadtimes dynamically, it does not take into account the mix of products, it fails to estimate the impact of setup times and changeovers on the overall capacity of resources and subcontractors, it lacks pegging of orders and it cannot do attribute based planning other than at the finished goods level. Consider this, if SAP-IBP fails to do full pegging of orders, how can it accurately determine the true cause of lateness of any orders? If it does not model capacity of resources properly with dynamic leadtimes, then how can it provide accurate ATP and CTP? -- There is something to think about next time you are wondering why your inventory and service levels are never on target.
Fact: realistic operations plans require accurate modeling of the supply side and all-in-one unified planning environment (what Gartner refers to as the ultimate or stage 5 supply chain). Transformation of plans into execution requires “model integration” By model integration, I mean the ability for S&OE engine to have the same data model as the S&OP albeit at different levels of detail. Accurate plans remove the need for excessive what-if scenarios and wasting time to adjust the plans, over and over. Accurate plans ensure delivery performance levels that are required by customer, by region, by product, and so on. Accurate plans provide much better visibility into the supply chain and what is possible and what is not. Accurate plans ensure the financial projections are realistic and reliable, as you plan for the long term. As an example, one of Adexa’s clients, a global high tech manufacturer plans tens of millions of orders within minutes with accuracy of almost 100%. This is done for both long term planning, as well as short term changes in the orders. Reallocation takes place every day depending on changes in the orders and availability of resources. Visibility is provided for close to one hundred sites worldwide. The best part is that planners have no need to make any adjustments and changes once the plan is published, i.e. they can go home early!
The bottom line, unless the plans are accurate and the model of the supply side is a true reflection of your supply chain then you are not getting much more than just half-accurate, but pretty reports. Spreadsheets will do just fine at this level of planning and give you a lot more control too, not to mention they are most certainly a lot less expensive! By deploying S&OE solutions to SAP solutions, as experienced by more than 70% of our clients, you will experience enormous benefits of what true plans can do for your organization: lower cost of operations and much improvement in delivery performance across all customers and products.
I encourage you to go beyond what is peddled today as S&OP, by SAP or anybody else, and look how much further you can go with more accurate and realistic plans by considering technologies that transition your supply chains into S&OE.
Planning is all about risk mitigation. Risks come in many forms: too much inventory, too little inventory, not enough to meet the demand, delivery interruptions because of weather, supplier issues, Acts of God, sanctions, labor disputes and so on. We also plan because we cannot react fast enough when the need arises. The more reactive we are the less planning is needed. How far in advance we plan has to do with how fast we can get what we want. Not every little detail needs to be planned. For example, when you plan for a road trip there is no need to plan for all the bumps on the road. The shock absorbers take care of that. Because they can react a lot faster and therefore remove the need for that level of planning. Furthermore, there is very little “cost” associated with running into small bumps on the road. However, if there is a potential snow storm on the day of travel then the cost might be a lot higher and re-planning needs to be done unless you have already accounted for this risk!
How can we do that? When the original plan was made, because of the season there was a chance of winter storm on that day of travel. We should have sufficient data (in our head for this situation) that indicates snow could be an issue during the winter season in certain parts of the country. Therefore, a risk factor needed to be attached to the plan for that reason. If the risk factor is high enough then a plan B is devised. The latter could be acquiring snow tires or taking the train amongst others. A summer travel would not have that risk factor but it might have other potential risks. Thus, plans and risks go hand in hand. With every plan, the associated risk needs to be taken into account to ensure that when and if the plan cannot be executed what the potential cost would be. Is there an alternative plan? Is it more expensive? And is the cost high enough to pay the premium to avoid the potential cost?
In supply chains, we are constantly facing these situations. However, we are not necessarily aware of all the underlying trends and moving parts that change the risk factors. For example, if the product mix changes, then our reliance on some suppliers become more than before. And if these suppliers are single-sourced and/or in earthquake zones then we could be facing a much higher risk than before the product mix was changed. If a new product is introduced and demand is much higher than expected, then do we have the additional capacity needed by the subcontractors to meet the surge in demand?
This is where planning systems become extremely valuable. A planning system has to be able to perform two tasks: Look for underlying changes in risk factors as plans are made for the future and identify the areas of brittleness. Secondly, when and if the risks are too high and the cost is justified, recommend alternative plans as to what can be done when and if the inevitable but unexpected occurs!
Performing such tasks is beyond the capability of human mind especially when it is done in an almost real-time manner. A note of caution, what we are describing here is not just a “what-if” analysis. This is having a system that is intelligent enough to evaluate the future risks and make changes to the plan and/or recommendation as to what needs to be done, based on the severity of the risks.
If I arrive on-time or early at the airport but there is a problem with the airplane and a long delay in the scheduled takeoff time, does it get me to my destination on time? Clearly the answer is No! Is it better for me to stay in the comfort of my home and go when I know the plane is ready to take off? Or even better, find another flight to my destination so that I get there on time. By going to the airport at the “wrong time” and waiting I am only increasing “WIP” or waiting time and I am also contributing to airport congestion which adds to the traffic and boarding of other flights possibly causing others to miss their flights. Such problems can be avoided by an intelligent planner or, release strategy, that can figure out exactly what the right time is for me to leave home given the traffic situation, the speed of cars, the parking time and time it takes to go through security. This kind of predictive planning is ideal for releasing lots in a semiconductor manufacturing line where depending on the mix of products, availability of resources and masks as well as WIP, it can decide which lots should be released and which ones should be held back so that we meet the following three objectives optimally:
- Cycle time
- Equipment utilization
- Delivery performance
The following diagram shows the relationship between these three parameters and how they change with WIP increase. In a high mix environment, increase in WIP does not necessarily imply additional wait times or delay in delivery because of multiple routes and balance of allocation of jobs by the system. The grey shaded area represents optimal region of operation where the desired objectives can be achieved.
In environments where there is a high mix of products, such as foundries, we can increase the number of lots released without increasing their waiting time by ensuring that they are balanced across different bottleneck equipment such as lithography equipment. Given the complexity of such environments where each process has 400-600 steps using hundreds of equipment requiring anything from 10 minutes to 10 hours with highly sensitive set up times (implanters) or batching requirements (ovens), one has to intelligently look ahead and look behind to ensure proper balance of lots re-entering the process and or entering the process with different priorities.
Unfortunately, sequencing engines with simplistic rules have been given too much attention in order to solve such a complex problem. Through years of R&D, we have concluded that unless a proper release strategy is deployed, sequencing would not be of much value. It is a reactive engine not a preventive one. But more importantly, in the presence of an adequate release strategy, sequencing can be a liability in the sense that it would try to resolve issues locally not being aware of the potential issues it might be causing 50 steps later! Can you imagine being at the gate, and the airline personnel try to sequence your entry into the plane when the plane is not even at the gate or being fixed!
One other myth is the use of simulation tools to plan fabs! Simulation tools look nice and show movement. It is like a video game, we all enjoy watching it. However, they DO NOT PROVIDE a strategy. They only show you where the problem might lie ahead without telling you how to avoid it. How could they? They do not look ahead; by definition simulation is one sequence at a time!
As in our opening example, a good release strategy is aware of the right mix of products in the fab as well as the work load of each equipment, now and future, and is constantly trying to balance what needs to go next such that the bottlenecks, as they are changing, will be fully utilized and at the same time keeping in mind which lots need to be ready and when for on-time delivery. In fact, our research shows that in the presence of a good release strategy, a simple FIFO is the best sequence for the resources. In the context of our airport example, if you left your home at the right time, as you approach your gate, without much waiting, you will show your boarding pass and get into your seat for takeoff. No need to be sequenced!
Topics: Supply Chain, Supply Chain Planning, Supply Chain Performance Management, Manufacturing Software, Manufacturing Planning, Inventory Optimization, Semiconductor, Factory Planning, Fabrication planning
Most planning systems deploy more of a reactive strategy than a predictive one. In the former category, when a problem is identified regarding a capacity shortage, material shortage, or arrival of a high priority order, the system addresses the issue by rearranging the plan. There is nothing wrong with this except that it is not optimal and it is a Band-Aid solution that could have been avoided in the first place! Let me illustrate this using a simple example. If you release too many jobs to your resources (or factory), you build a big queue in front of resources creating much WIP adding to your inventory and delaying delivery dates. In addition, you diminish your capability to address inevitable but unforeseen problems such as shortages, equipment breakdowns, and arrival of surprise orders. But the interesting part is that, most people try to resolve this issue by having better sequencing rules for each resource. As you can see a problem that could have been avoided was created and now we are trying to reactively resolve it by locally expediting, which is almost impossible. This type of strategy is prominent in many S&OP solutions which tend to operate at a high level not knowing how the plan can be executed. They assume a fixed lead-time and assume maybe ONE bottleneck resource for the entire plan for every site, and then expect to have an accurate plan! However, when it comes to executing such plans there will be a lot of expediting and adding shifts and delays to name a few. Such an approach is no different than the use of spreadsheets for planning and use of fixed lead-times, which really implies infinite capacity planning. I remember MRP systems did that really well! Are we back to the technology of 80’s?
We believe that predictive planning avoids the problems in the first place and diminishes the need for reactive solutions. We also believe that any plan generated by the system has to be accurate enough so that it is executable. Spreadsheet type of planning (pre-defined bottlenecks, fixed lead-times and bucketed capacity) deployed by most S&OP solutions are simply NOT accurate enough! They give you a false sense of hope and control as well as poor visibility into what can be accomplished; resulting in erroneous delivery dates. By performing predictive planning, you can account for potential issues of shortages and breakdowns or even quality issues in advance. Furthermore, one can “release” the orders (virtually in subcontractor facilities or actually in your own) in a way that they do not have to wait for a long time in front of resources, reducing WIP and at the same time maximizing utilization. By producing realistic plans based on “shifting bottlenecks,” one can also ensure realistic due dates maximizing on-time delivery. Predictive planning is performed by having an accurate model of your supply chain (yours and your suppliers’) and understanding the mix of products that need to be built and how they compete for resources in a dynamic manner. In addition, making sure that we account for probability of breakdowns, demand variations, maintenance schedules, supplier lead-time variations and so on. By taking into account all such variations, one has a realistic model of the supply chain and can precisely predict the behavior and how each order can be delivered through different choices of the Supply, Make and Distribution. The key is to take all such combinations into account and optimize the use of resources and inventory using an optimal order release strategy that maximizes delivery performance, minimizes inventory and optimizes the utilization of resources. For additional information, please refer to Adexa Whitepapers on this topic by clicking HERE.
Planning is the world’s second oldest profession! Why? Because no matter what you want to do, you need to plan for it. You may argue that there are things that you do not plan for: accidents, Acts of God, inventions. However even these are somewhat planned for in case they happen. Buying insurance is a good example of such planning. Perhaps emotions and feelings are exceptions! But even these were planned by nature for our survival. In the world of manufacturing and supply chain, planning is probably the most important aspect of the business since it has to do with getting the right product in the hands of the consumer on time in the right place. In the absence of proper planning the cost can be very high leading to the demise of the business. Despite, the importance of planning, it is not enough to ensure on-time delivery of goods at the lowest cost. This is due to the stochastic nature of events that can change the demand or supply. Examples are a breakdown of an equipment or plant due to earthquake. Or, sudden increase in demand or shortage of supply from a supplier can lend the original plan somewhat under-optimized. Hence the title of this piece which is a quote from General Eisenhower.
We conclude that S&OP is good for aggregate level decisions for things like expected total production required in a month, but not good enough to figure out how much to produce of each individual product along with the capacity needed of each machine type. You need to have accurate plans that can be executed; and while they are being executed the plan is constantly adjusted to account for the changes that were not foreseen by the plan. To be able to do this, one needs S&OE (Sales & Operation Execution) system that can translate the original plan into more and more detail. A typical S&OP plan can only be as accurate as 40-65%. When combined with S&OE, the accuracy can be as high as 98% or better. We have actually performed simulation studies that confirm the above results.
What S&OE does is simply take the constraints defined by S&OP and apply additional detailed constraints in order to make the plan work and more accurate. In this process, it enables the users to see the details of what will be delivered and where and what potential issues need to be addressed, if any. For example, the S&OP Plan may say the forecast for the month of Product group A is 100, while S&OE says how much for A1 and A2 per day. S&OP says 80 hours of drilling is required, while S&OE says how much of each type of drill is required each week.
Adexa solutions deploy attributes to define the characteristics of machines, processes customers and suppliers, in order to mold the solution to a particular environment. Furthermore, as the supply chain changes and business or business priorities change then attributes are used to adjust to the new conditions. This has been extensively covered in a number of white papers that can be referenced by clicking HERE. In this paper, we discuss the use of attributes for a greener supply chain and how they may be deployed to improve on the carbon footprint and use of hazardous materials.
Consider a process or an equipment that has more CO2 emission than other alternatives. The value, or relative value of the CO2 emission becomes one of the attributes of that process. On the other hand, cost might be another attribute of the process in a way that the lower cost process may contribute more to CO2 emission. Attributes, once defined, form dynamic constraints that are used by the system’s search strategy to find the optimal solution. The optimal solution can vary depending on the region, government regulations, carbon use quotas and even customer. Within the framework of such restrictions as defined for the attributes, the system can recommend the highest use of lower cost process without going over the CO2 emission limits as decided by the company policy. As alternatives, the system may recommend higher cost processes, use of substitute materials or manufacturing in a different region.
Another example, is the use of different transportation means in order to deliver gas or oil to different regions in Europe. The choices are by sea or land as well as pipelines. Each has attributes of cost, carbon footprint (depending on the distance travelled) and of course time to destination as well as associated risks. For example, it is imperative that oil gets from A to B if it is to be used for heating in the middle of North European winter. Land delivery has a much higher risk of road closures in certain regions.
As a final example, some customers or retailers may have a preference to sourcing products that are eco-friendlier. This kind of requirement is sent to the suppliers as an attribute that is then built into the BOM of the product. On the manufacturing side, the use of such embedded materials in the product is treated as an attribute of the material as well as the product. This attribute is then taken into account and honored as a constraint when plans are made by the system for making and delivery into those customers.
As it can be seen, attributes can be used in order to define local and global constraints on the operations of the entire supply chain including tariffs, carbon usage, supplier types, transportation means and material properties depending on the regions, government/international regulations, and company policy. For instance, a company can reduce the global value of an attribute (say, CO2 emission) by 10% per year by region and product. The system would then plan production of the entire supply chain by making sure that the local attributes used in every region does not exceed the maximum defined by the management of the company. We welcome your feedback and sharing innovative uses of attributes in your supply chain operations.
There is risk in almost everything we do. It is unavoidable. Supply chains are no exception facing all kinds of unexpected but inevitable surprises. Some can be very costly to the company. It is imperative that the management are prepared to deal with unfavorable issues when they occur without building too much redundancy increasing the cost of operations. In a typical supply chain, having thousands of SKU’s and suppliers as well as other factors such as geopolitical issues, labor related issues and demand volatility, makes the supply chain operation very complex and in the absence of appropriate tools almost impossible to manage in an efficient manner. The key is to identify the potential risks before they happen so that adequate measures can be put in place.
There are many ways to assess risk vs cost and reward. As an example, one can use Multi Echelon Inventory Optimization (MEIO) to assess risk of on-time delivery vs cost. This can be done by SKU and customer. For certain customers, the desired delivery performance must remain at 98% or higher. Obviously this can be accomplished at a higher cost of inventory at different stages of the supply chains. On the other hand, for many other customers, a delivery performance of 90% might be acceptable at much lower cost of operations. As the demand patterns change, MEIO behaves as an almost perfect postponement strategy, to show where and when inventory is needed for a desired delivery performance and cost by customer and SKU. This algorithmic approach, based on probability distribution and queuing theory, is by far superior to the traditional methods of historical data such as moving averages and/or min-max types of approach.
Having visibility into meeting the financial goals of the company is critical. Any risks associated with that must be detected as early as possible and addressed. Likewise, meeting delivery performance for certain key customers, making sure that the right mix of inventory is available to keep the production running, knowing what options are available in case of capacity shortage, or material running out (or not delivered in time) are all factors that may increase delivery risks, increase cost and even cause loss of market share. Optimization models of systems designed to assess the impact of risks can act as a crystal ball to provide visibility to the end users and furthermore provide guidelines and advise end users as to what the best course of action would be. It is a proactive way of responding to potential risks than reactive.
One other critical use of systems is to perform what-if stress tests on the entire supply chain. By either overloading the supply chain model or trying to break certain links in the chain, one can observe the consequences of such events and what can go wrong, what the financial impact would be and what can be done from the convenience of your desk, before it happens! Preventing such potential disasters are how modern heroes are made of in the world of leading companies! Learn more about Supply Chain Risk Resiliency by clicking this link.
Where is the “E” in S&OP?
According to Gartner, there can be no effective S&OP process without an S&OE—Sales and Operations Execution, process. In other words, why make a plan if cannot be executed accurately or cannot be translated into execution? Here is what Gartner says:
• The No. 1 challenge among manufacturing companies is connecting sales and operations planning (S&OP) to operational plan/execution.
• Value-adding, effective S&OP process cannot exist without S&OE, as it provides the planning interface to execution.
There has been a recent surge of interest in S&OP. The attraction is primarily in integrating sale and operation as well as better visibility into potential issues. However, the major problem with most, if not all, S&OP systems is that they are far from accurate especially when it comes to capacity and mix of products as well as order level pegging. The inaccuracy of plans leads to all kinds chaotic manual adjustments and tweaking just to get it to work. S&OP shows you the direction but cannot take you there! In order to arrive at your destination at the right time and place you need to be able to execute the plan not just get a high level idea of which direction to go. There may be traffic jams, there may be road closings, there may be flat tires; and there may be bad weather and bad road conditions. In all such instances your S&OP plan is inadequate to deal with such inevitable but unpredictable issues and cannot give you any help as to what your best course of action should be. Their best suggestion is “keep simulating and perform What-if” all possible conditions that might occur! This is not a feasible approach!
To be able to perform execution, you need to have:
1- Accurate plans (not just rate-based planning like spreadsheets)
2- Have a unified data model between plan and execution engines
3- Ability to adjust the plan as needed
Furthermore, execution systems, just like vehicles, can handle all the bumps on the road to make the ride a lot easier. Shock absorbers in your vehicles are precisely a method of execution of plans. In their absence, the ride can be extremely uncomfortable. This is often the case when execution systems are missing and people often end up performing all kinds of expediting and manual changes, trying to absorb the “shocks.” Even more importantly, the main issue is that the financial predictions that they had made regarding the plan at the S&OP level is no longer valid and totally false. For example, use of more expensive substitute materials or more expensive freight methods, because of late deliveries, can substantially add to the real cost of products. To this end, the S&OE system needs to have the capability to monitor the financials, as such changes are made, on an on-going basis. Costs based on S&OP alone is wishful thinking and have no resemblance to the actual.
Another important point that was made earlier above was having the ability to translate plans into execution. In other words, a seamless transition between planning into execution. A unified data model in plan and execution systems is a requirement for this to happen. Furthermore, the planning engine must have a realistic model of the actual environment so that the plans are produced in an accurate manner. Current methods of capacity planning which are based on bucketed rate-per-period, is very inaccurate and insensitive to the mix of products. Even if you had two products A and B, if B takes twice as long as A to make, one cannot plan this unless you know how many B’s and how many A’s are needed. An average of X per day or week is very misleading which is exactly what almost all S&OP systems do. In many cases, such as asset-based industries, the actual models of equipment are a must in order to produce accurate plans. Capacity depends on set-up times, batching, availability of tools to be used on the equipment, processing times and so on.
Lastly, almost all S&OP systems cannot trace orders to the components and supplies that are to be used for every order. Thus, the ability to peg orders, to qualify suppliers, and to find root causes of issues and latenesses are totally impossible for these systems! They can only say what is short and which orders will get what is available. In high tech, pharmaceutical and many other industries, pegging to the right supply, to the right supplier and to the right processes are requirements. S&OP systems must be able to demonstrate such capabilities or else the actual execution of the orders would be a nightmare.
S&OP and S&OE are two essential processes that go hand in hand and in the absence of one or the other the whole purpose of having the system is defeated.
For further information on how Adexa’s S&OP and S&OE work to avoid the aforementioned issues, please send an email to firstname.lastname@example.org or visit our web site: www.adexa.com
In a few recent sessions with industry analysts, we were surprised that we were asked if our software is in-memory computing! Given the fact that for over 20 years we designed our applications to have all the data in-memory for computation, our immediate response was: Is there any other way of doing it? The response was, yes, there are others which bring in the data from the database when they need it but now they are changing and they are getting orders of magnitude improvement in speed! This improvement in speed must have caught the attention of the analysts which brings us to the core subject of this article. There is more to speed of application than just having all the data in memory. The latter is the easy part. There are also some vendors, try to improve speed by abstraction and over-simplification. I am sure you are aware of quite few who deploy “Spread-sheet” type of capacity planning in their S&OP applications. That is forming weekly or monthly buckets with fixed lead-times! This approach typically either dumbs down how to deal with capacity, or ignores it altogether. It is the old method, with NO notion of product mix and real processing time, that has been around for decades but with a new user interface which makes it slightly more attractive. Therefore, any gain in speed is offset by a very inaccurate and unrealistic plan. In addition, it has no order level information OR any order level pegging functionality. You might as well use your spreadsheets since they give you even more control!
To gain real improvement in speed with proper representation of capacity of resources and equipment, deep modeling capability is needed and the mix of products must be taken into account. In addition, to IMC, one needs to have data representations and algorithms that provide real-time answers to very complex supply chains at order level. As an example, if one material is not available, does the system go back to search all over again for a new method of making or will it just backtrack one step to find an immediate substitute pegged to that order? If a resource is a bottleneck, will it look for a whole new routing or will it look for an alternative, process or equipment. How this data is represented and how the algorithms divide and conquer in parallel processing is what makes the application fast. Just using IMC is only the beginning, there is a lot more that goes into a comprehensive planning system that can analyze tens of millions of data points from material availability to resources and tools and skill levels, to say a few, in almost real-time.