In 1994, the IBM Micro-electronics Division, itself a fortune 100 size firm, put in place a major effort to create best in class supply chain planning process and software including demand planning(DM), central planning, available to promise, et al. I was fortunate to be an original member and had the opportunity to work extensively on all key components including DM (1996-97) – created an estimate of demand – a forecast. With 20 years of experience in computational methods at that time, I thought DM would be a walk in the park. I quickly learned it was illusively complex.
In the simplest inventory situation, the only variability is in the quantity of demand for a single day. There is no trend up or down or seasonal effect. The demand today is independent of the demand for tomorrow. Additionally, we will assume replenishment time is zero. That is when we place an order for additional material it arrives immediately – sort of like we have a Star Trek transporter. However, we can only reorder every N days, where N might be 3, 4, 20, etc.
Most folks involved in Sales and Operations Planning (S&OP) for supply chain management have heard the terms “optimization” or “linear programming” with regards to supply planning and have cringed at the sound. Over the past few years a new “cringe” worthy term has emerged – “machine learning” which is sometimes used with the term predictive analytics or data mining. The purpose of the material below is not to explain optimization or machine learning, but to provide an easy to follow example from numerical methods applied to high school algebra to illustrate the key computational principle that supports important decision technologies. We will see it as Generate,Test,Next.
The term ‘optimization’ can and does have different meanings to different groups. For the folks who build and develop scheduling algorithms, creating the best schedule is defined in terms of cost criteria – perfectly logical. For business settings (from manufacturing to hospitals) optimization refers to the entire process. Let’s look at the scheduler’s world to identify key steps to optimize the process.
Each year I work with new bright-eyed future experts in planning and scheduling as they make the transition from their academic studies to the murky world of applied planning and scheduling. One of the first rules of thumb I suggest is to ensure everyone has the same view of the problem. Is this a planning problem or a scheduling problem or both? If both are being done simultaneously, the following example quickly illustrates the value in separating them.
One value of supply chain modeling is the ability to explicitly model capacity or constrained resources. Which products get what capacity in what amount at what time? How does this impact on-time delivery? Where do I add capacity? These types of questions are custom made for a “little bit of math” and difficult to do in a simple spreadsheet model.
Your demand planning process is complete and demand statement is created; excellent, but the sales and operation planning journey is far from complete – the next critical step is matching or balancing assets (capacity, materials, people and projected supply of finished goods) with demand to answer three questions that are part of being a responsive organization: