Demand-driven Versus Traditional Materials Requirement Planning
Written By: Predrag Jakovljevic
Published On: August 8 2005
The Realities of Manufacturing Today
Nowadays, manufacturers are increasingly subject to massive pressures to drive down costs and increase efficiency. However, these pressures often invalidate the traditional materials requirements planning (MRP) batch-based manufacturing planning and product costing approaches. Moreover, companies struggling to serve their customers using purely traditional MRP methodology are often unable to meet the demands for agility and responsiveness that consumers at the end of the supply chain are requesting. To make things worse, with product life cycles decreasing, it means that manufacturing and distribution are increasing in complexity. For the manufacturer, this translates into a need to better manage customer demands and expectations, and to respond accordingly.
With ever shorter product life cycles in fashion due to shifts in buying trends and marketers attempting to best guess what fickle consumers will desire from season to season, the supply chain cannot afford items that sit on shelves (tying up capital and facing obsolescence). Whether these items are finished products, components, or subassemblies becomes irrelevant in this scenario. Therefore, there is a real need to reduce inventory throughout the supply chain nodes. For example, carmakers tend to see an increase in supplying customized products, with consumers able to specify everything from the color pallet for bodywork and interiors to choosing the latest electronic gadgetry for the dashboard. One can even witness customization creeping into mass-produced goods too, where certain versions of items of wide ranging types are being made available through different outlets—may it be food, fast moving consumer goods (FMCG), etc.
Needless to say, the need to reduce capital employed within the manufacturing enterprise and the trend of outsourcing manufacturing to lower-cost regions overseas, which typically increases lead times (which customers do not appreciate), only complicates the conundrum of low costs and increased efficiency for embattled manufacturers. This means that customer management has to move up several steps of the efficiency ladder, with the best companies staying very close to their customers.
In reality, the way to overcome these difficulties and better serve customers might often be to adopt demand-driven manufacturing principles. For example, manufacturers could make product as close to the point of order as possible, anticipating needs, and delivering within an acceptable time frame. Demand-driven manufacturing is not a new concept. Japanese manufacturers went down this route back in the 1980s. Now the rest of the world is slowly waking up and moving away from batch production towards demand-driven manufacturing, because wavering demands from customers create too many demands on the inflexibility of traditional manufacturing methods.
It is old news that optimizing within the four walls of the factory is no longer a workable solution, and while outsourcing may be seen as important in lowering the price of finished goods, it causes further problems by increasing lead times in a world where decreasing lead times are equally necessary to satisfy the customer.
Further, manufacturing anything—from mobile phones and computers, to cars and toys in the discrete manufacturing segments, to meat processing, producing paints, and brewing beer in the process manufacturing segments—can entail extremely complex business processes. Namely, parts or ingredients are needed to make components, which in turn, need to be configured or assembled before a final product can be delivered to a customer.
This is Part One of a two-part tutorial.
Part Two will discuss demand-driven planning.
Scheduling and Forecasting
Alternatively, the manufacturing process itself can be rather straightforward, but is subject to difficult scheduling requirements, due to long lead times and fluctuating market demand. The APICS Dictionary defines demand as a need for a particular product or component, which could come from any number of sources. This includes customer order or forecast, an interplant requirement, or a request from a branch warehouse for a service part or to manufacture another product. At the finished goods level, demand data are usually different from sales data because demand does not necessarily result in sales. For example, in the sales scenario, if there is no stock, there will be no sale. Generally, there are up to four components of demand: cyclical component, random component, seasonal component, and trend component.
On the other hand, demand management is the function of recognizing all the demands for goods and services to support the market place, which involves prioritizing demand when supply is lacking. Proper demand management facilitates the planning and use of resources for profitable business results. In other words, it integrates supply and demand information so as to optimize operations. Forecasting applications, which predict activity over a weekly, monthly or even yearly time horizon, remain central, but demand management is a broader activity that can include replenishment, sales and operations planning (S&OP), integration with marketing, order, and customer resource management (CRM) systems.
The traditional time-series forecasting approach the averages of the past performance of a demand stream to anticipate further demand, but more sophisticated systems take into account factors beyond historical demand, employing statistical methods to remove biases. These more complex forms of forecasting determine and predict the effect of "causal" or "event-driven factors," and macro-economic indicators, might have on demand. To that end, recent advances in forecasting have focused on gauging the impact of pricing and promotions, product introduction and obsolescence, intermittent demand, and product proliferation, while forecasting accuracy is improved through collaborative processes that allow sales and distribution channels to work interactively with forecasters.
Either way, planning and controlling this enormous flow of processes and information requires sophisticated software. Adding to this complexity is the distribution of manufactured goods to market, since many variables come into play, such as lead-times; customer orders; internal orders and inventories of products; components; and raw materials. Many decisions must be made, such as when to re-order components or parts, how much inventory to keep, and so on.
It was not until the late 1970s, when computers begun being used in the manufacturing process, that some of these complexities could be mitigated. Solutions such as re-order point (ROP) systems and MRP have been the most common tools to plan for when, and how much, of a certain component or part should be re-ordered. To refresh our memory, a ROP system is an inventory method that places an order for a lot, whenever the quantity on hand is reduced to a predetermined level, known as the reorder point. On the other hand, master production schedule (MPS) is the anticipated build schedule for those items assigned to the master scheduler. It is a set of planning numbers that drives MRP, and it represents what the company plans to produce, expressed in specific configurations, quantities, and dates.
Finally, MRP is a set of techniques that uses bill of material (BOM) data, inventory data, and MPS to calculate requirements for materials, to make recommendations to release replenishment orders for materials. Further, because it is time-phased, it makes recommendations to reschedule open orders when due dates and need dates are not in phase. For more definitions, see Glossary of Enterprise Applications Terminology.
Impact of Computer on Planning Process
The impact that the computer had on material planning and enterprise management in the 1970s was immense. From manual planning to the huge inventory of posting card decks, this new computerized system promised to automatically plan, build, and purchase requirements based on the finished products to be shipped, the current inventory on hand, the allocated inventory for other orders, and the expected arrivals. The posting, originally done manually on input/output cards, was replaced by transactions directly made in the computer and documented on pick lists. The amount of inventory was supposedly visible to anyone with access to a computer, and did not require the user to go to the card deck.
Hence, MRP represented a huge step forward in the planning process. For the first time, based on a schedule of what was going to be produced, which was supported by a list of parts that were needed for that finished item, the computer could calculate the total need and compare it to what was already on hand or committed to arrive. This comparison could then suggest an activity to place an order; cancel orders that were already placed; or simply move the timing to expedite or delay existing orders. The real significance of MRP was that, for the first time, the planner was able to answer the questions "what?", "when?", and "how much?". In other words, rather than being reactive and waiting until the shortage occurred, the planner could be proactive and time phase orders, including releasing orders with multiple deliveries. Indeed, the enterprise systems currently in use by most large corporations worldwide are an evolution of the MRP systems, which had managed to regiment former chaotic manual systems, to a degree.
Nevertheless, some simplifying assumptions were needed to allow the computers of the day to make the required calculations. One was that the orders should be started at the latest possible date to provide for minimal inventory while still serving the customer's need on time. This method is referred to as backward scheduling. Therefore, all orders were scheduled backwards from the desired completion date to calculate the required start date.
Also, there was no inherent slack time in the schedule. The downside of this assumption was that if there were any hiccups in the execution of the plan, the order would most likely be sent to the customer late. Further, if only one part required for the finished product was going to be late, there was no automatic way to know the impact on the other needed parts.
The result of the MRP run, which can take several hours in some environments, is supposed to tell planners how to organize their work by releasing production orders. MRP will, by default, create orders with specific due dates for products and when they need to be manufactured. Companies prioritize resources based on these calculated due dates. The unfortunate result is that other orders, perhaps more important, are neglected, which often leads to overtime in the factory. Therefore, slack would often have to be built into the schedule through conservative, often unjustifiably pessimistic lead times. Despite this drawback, the benefits of the system far outweighed the costs and more companies began to embrace the tools and techniques of MRP. For more information, see Enterprise Applications—The Genesis and Future, Revisited.
Combined with information from actual customer orders, MRP is still the most widely used tool in manufacturing industries to track, monitor, and order the volumes of components needed to make a certain product. However, for these reasons, many manufacturing environments have discovered that MRP has trouble controlling stock levels, which results in poor delivery performance. Also, MRP is incapable of handling demand-driven, ever-changing manufacturing, working better when demand for a particular product is constant (fairly even) and predictable. If there is any variation, however, then MRP loses many of its advantages and the benefits of using alternative planning approaches increase.
In fact, the main flaw with MRP is that it is too deterministic. It is too rigid, as it does not allow for natural variations that occur in real life, such as people getting sick or going on strike, truck and shipment delays, machines malfunctioning, quality issues requiring scrap or rework, and customers not order according to forecasts. In other words, MRP is a static model of a stochastic reality, which happens to manufacturing all the time, and is based on customer orders, available parts, and so on. MRP attempts to apply a high degree of precision to something that is inherently imprecise.
Again, MRP is a system that strives to plan replenishment just before a withdrawal from stock, which does not work in some manufacturing environments. In the language of logistics experts, MRP is a "push" system, which schedules production based on forecasts and customer orders, and thus creates plans to push materials through the production process based on forecasts that, by nature, are not accurate. That is to say, traditional MRP methods rely on the movement of materials through functionally-oriented work centers or production lines, and are designed to maximize efficiencies and lower unit cost by producing products in large lots. Production is planned, scheduled, and managed to meet a combination of actual and forecast demand, and thus, production orders stemming from the master production schedule (MPS) and MRP planned orders are "pushed" out to the factory floor and in stock. External suppliers also work to support planned production, while materials management often relies on maintaining sufficient inventory, using a make-to-stock (MTS) rather than make-to-order (MTO) or assemble-to-order (ATO) approach.
This concludes Part One of a two-part tutorial.
Part Two will discuss demand-driven planning.