What MRP was missing for APS to emerge
Many people ask us what APS software is, what it does and for what kind of demand it's relevant. So in this article we'll discuss a bit more about how APS software emerged and the main gaps in MRP and ERP systems it fills.
THE ORIGIN
Production planning, in its broadest concept, has always been at the heart of industry. Maybe in recent decades it hasn't been treated with the deserved attention compared to other areas, but it plays the protagonist role in the history of management technology for industry.In the 1950s, computational development introduced CAD (Computer-Aided Design) to Engineering, but its high cost made it accessible to few — namely the aerospace and automotive industries, which had (and still have) very high technical demand. But management was still “by hand.” The growing economic expansion, alongside industrial development, made planning challenges more critical. As a result, the first business-systems initiative for industry was not a financial or commercial system, but rather MRP (Material Requirements Planning), created in the 1960s by Oliver Wight and Joseph Orlicky precisely to plan the needs of materials to manufacture and buy.
For those joining mid-conversation: basically, MRP generates net needs of these materials from finished-good needs and the explosion of those needs through the different levels of a product structure, generating each material's needs at each level, whether an internally produced intermediate component or a purchased raw material. Each of these needs has a date set by standard replenishment lead times for each item. MRP is nothing more than a series of simple, chained calculations — but in industries with many components and structure levels, that's a huge help, especially considering the era it emerged in.
Since innovation loves problems, growth in demand for industrialized products after the 1960s ran into the next planning challenge: capacity. It wasn't enough to know which materials would be needed in each period, but also whether the indicated volume could be produced or purchased, and to see the financial impacts.
So in the 1980s, MRPII emerged. In it, resource management started being considered, with interfaces to the financial and engineering areas to understand physical and financial needs of labor, machinery and materials to execute a production plan.
From MRPII the Master Production Schedule (MPS) is defined — the definition of finished products to be produced in a period (already disaggregated by SKU). So, knowing what we'll need in finished goods and roughly understanding it's possible to produce, we explode the needs of all materials through the already known MRP. Finally, all needs generated by MRP are validated. It walks through all the operations registered in the product's manufacturing routing and consumes the times needed to produce those products from the total capacity of each productive center. At the base, it's what many know and practice as machine-loading in industry.None of these concepts came ready with the birth of MRPII. They were refined over time. Growing industrial demand and this technological evolution spreading to several areas still lacking allowed Integrated Management Systems — or simply ERP (Enterprise Resources Planning) — to emerge in the mid-1990s. They integrate the different areas of a company: accounting, human resources, finance, engineering, manufacturing, sales, among others. ERPs emerged trying to incorporate one or more of the functions described earlier.
The limitations of the systems existing up to that moment for accurately performing production scheduling encouraged the creation, in that same 1990s decade, of the concept of Finite Capacity Scheduling (FCS), which later evolved into Advanced Planning and Scheduling (APS). It's vital to understand that both emerged to fill the deficiency of the other systems in managing finite production capacity, as well as queue, routing and synchronization management between production operations. Let's explore these limitations:
Fixed Lead Time
Like most people in the world, you probably commute daily from home to work. Even though you have an estimated time for the commute in mind, if you monitor that time daily, you'll notice significant variation — especially if you live in a big city. That time can be even more volatile if there's a construction project or event on a street in your route causing you to vary the path. So a factory is a megalopolis that every day has more or less congested routes. For you, leaving 10 minutes early to make sure you arrive on time may cost little. But for an industry, those “10 minutes” at the scale of hundreds or thousands of Production Orders, deadlines and resources generate a very costly error from a planning standpoint. Both MRP and MRPII use fixed lead times — that is, they don't see that your work commute changes duration if you alter your route or if streets are congested. The more operations per product, the more resource or routing alternatives, and the larger the demand variation over time, the larger this divergence will be. The direct consequence is the difficulty industries have setting delivery dates and maintaining inventories at healthy levels. It's no coincidence that a large share of companies still relies on these systems and, not coincidentally, make up a growing volume of industries complaining about delays, stockouts and overstock.
Batch or Lot Processes
Another point of attention is batch or lot processes, such as thermal treatments, paint booths, galvanization, dyeing, wear machining and many other types of operations that process multiple products simultaneously. MRP and MRPII don't distinguish these processes from those defined by time per item (or weight, or length) or rate per hour. So these processes get poorly sized loads and the defined production plan will have less chance of being properly executed. Some claim these processes aren't always bottlenecks in their industries. That tends to occur in thermal treatments at metal-mechanical industries, but in practically all other cases these processes have a strong tendency to become bottlenecks quite frequently. Furthermore, based on the hundreds of industries we've visited over many years, we can say roughly half of them have this kind of process and therefore this issue.
Synchronization
When there's more than one operation to transform a product, there's naturally a sequence to follow between them. You can't pack the product before painting or assembling it. The Manufacturing Routing is the reference for this order between activities. But CRP doesn't handle this synchronization aspect well. It usually deals with it in two ways.In the first, it loads all operations in the same capacity period. However, depending on the queue forming at each process, a downstream process should have been allocated in the next period and its expected delivery date should have changed — what happens is the Production Order is late. In other cases, especially when process times are longer or there are many operations, you define that a set of operations will be in period 1 while another will be in period 2, and so on. A sequential logic over time is created, which is good, but the usual consequence of this staggering ends up being a great dilution of production. Total production lead times grow, the real value-add time over this total becomes low and work-in-process inventories consequently rise. This second scenario tends to generate fewer delays but at the cost of inefficiency created in the system.
Finite Capacity and Constraints
Perhaps this is the main factor that makes the MRP result just an answer key for an exam from which you can't demand a minimum grade because you don't know whether it can be solved. Defining what to produce or buy without validating capacity, in theory, wouldn't let us hold the factory accountable for coherent results. Usually there isn't just one bucket with capacity filling up: it's not only a machine or a workstation that constrains capacity. The people operating the machines, the tooling used, the physical space between sectors and several other criteria are also normally restrictive. A specific capacity analysis just for the main constraint of each process — or worse, just for the bottleneck process — is very limited.Now, if we tie this finite-capacity factor with its constraints to what we discussed earlier — synchronization — we'll see that the hole goes deeper. There's no point in doing a machine-loading to limit capacity if we don't see that the product that seemingly is promised to be produced and delivered actually can't be processed today because an essential tool for its production is being used on another machine. This kind of situation, involving tooling and labor, is more frequent than you'd think.
Sequence
If we look at the 7 Lean Manufacturing wastes brought by Taiichi Ohno, we see two of them are directly related to production sequencing: waiting and processing wastes (the latter through setups). Some processes have setup times that can reach 50% of total available time in certain periods. Ask specialists how long it takes to change a cylinder on a loom or to wash a tank before producing allergen-containing food. Depending on the operation sequence you set for each resource, you can significantly reduce mold and tooling adjustment times, tank and pipe cleanings, color matching and even operator movement. Since these systems (MRPII/CRP) don't consider the operation sequence, they tend to use average setup times — which can vary between 5 minutes and 3 hours — and using an average will likely cause problems in process end estimates. Since there are usually multiple operations to produce something, this seemingly small error reflects on downstream processes and a domino effect occurs, distorting the initial plan. In the same line of thought, if we don't dictate a sequence, a shift may produce a given output and leave for the next shift products that can't be processed by it because there's no team to perform the setup in that specific shift, or because it's a product needing more operators per machine and that shift has insufficient people, among other possible factors. This will generate more wait times between processes, the other Ohno waste. In short, in practice we see that systems ignoring important factors for accuracy fuel a vicious cycle of targets and empirical controls full of simplifications that hinder the connection between actions and results. To eliminate these gaps, we need a more advanced method and technology. Enter APS.
So, after all, what is APS software?
APS stands for advanced planning and scheduling — software for advanced production planning and scheduling. APS is a specialist system; its idea isn't to replace the ERP, but to complement it and address its gaps, working connected to it, processing production and sequencing information while taking into account all constraints, finite production capacity and the necessary synchronization, showing the best way to perform production according to each company's strategies. NEO is the largest APS consultancy in Latin America. Sequence your production with someone who specializes in technological solutions.