International Conference on Industrial Engineering and Systems Management IESM 2007 May 30 - June 2, 2007 BEIJING - CHINA Cover-Time Planning, An alternative to Material Requirements Planning; with customer order production abilities and restricted Work-In-Process * Anders SEGERSTEDT a, b a Industrial logistics, Luleå University of Technology, Sweden b IBDK, Narvik University Collage, Norway (part time Professor II) Abstract An alternative to the more commonly known Material Requirements Planning (MRP) is presented. The technique is based on a coverage analysis and is called Cover-Time Planning (CTP); CTP is in fact a reorder-point system, but with time instead of quantity as the decision variable. Similar to Kanban CTP restricts the total amount of an item but in the CONWIP way, not in every machine, but the total amount in stock and in the route from its components. Contrary to traditional reorder-point systems CTP uses the information in the bills of material to present forecasts for all possible component items. CTP shows inherent priority numbers and like MRP CTP makes it possible to estimate future work load. The main principle behind CTP is to wait as long as possible before production starts; to wait so that it is more exactly known what the customer asks for and by that avoid to tie up material and components in something that the customer will not ask for. Key words: Manufacturing, Supply Chain Management, Material Requirements Planning, Cover-Time Planning, Customer Driven Production 1 Cover-Time Planning Cover-Time Planning is an alternative to the more commonly known Material Requirements Planning (MRP) in systems for Enterprise Resource Planning (ERP). They are techniques which are required to presents signals for starting production before the final customer order arrives. Most manufacturers have to start some processes of their production in advance in order to present competitive delivery times and avoid delivery times which are as long as the total lead-time of the product. Some companies even do their whole production based on forecasts and have the product in stock when it is requested by the customer. Cover-Time Planning (CTP) implies that expected demand rates for end items, or items on suitable highest level in the bill of material, are transferred down (exploded) to underlying levels using the information in the bill of materials and reported lead-times, to calculate and estimate demand rates over a time horizon for all items. Each item is after that treated separately from all other items. For every item a current supply is defined: * This paper was not presented at any other revue. Corresponding author A. Segerstedt Tel. +46920491212. Mobile +46706444537 Email addresses: anders.segerstedt@ltu.se (Anders Segerstedt)
Supply = On hand inventory Backorders + Planned replenishments (1) The time this already activated or started supply will last to cover expected future demand is based on a coverage analysis and is called cover-time. If the demand rate is stable and the lead-time short, the cover-time can be estimated as: Cover-Time = Supply/Demand rate (2) If the lead-time is long and the demand rate varies a more accurate calculation of the cover-time has to be undertaken. Figure 1 illustrates an example where a more accurate calculation is needed. The figure intuitively shows how the cover-time is calculated. Figure 1. An example with varying lead-time. When for an item if: Cover-time < lead-time (incl. inspection interval) + possible buffer time (3) Available on hand inventory within lead-time < 0 (4) CTP creates a signal that a further replenishment is needed. Equation (3) is similar to the comparison between the inventory position and a reorder point. (The reorder point covers demand during the lead-time and the inspection time and contains a safety stock.) If in an application we have customer orders but no forecasted expected demand rate, equation (4) checks that enough replenishment orders are released to cover existing customer orders. Customer orders registered for delivery in the future ahead of the lead-time can be satisfied by future replenishment orders. 2 Customer order production Accomplish customer driven production is very difficult using MRP (Material Requirements Planning). If a lot of different variants of the end products are produced, it is hard to forecast when to start manufacturing and purchase of material before the final customer order arrives. Productions and purchases that have to start in advance in order to keep delivery times on competitive levels. To manage this while using MRP, awkward support systems such as fictitious planning structures are needed. Or you have to split products into different inventories or to work manually by deleting customer orders from
earlier forecasts, for further descriptions cf. [1]. Even though it is complicated, it never gets fine; it is easier to work with a traditional reorder point system. But, ordinary reorder point systems have other disadvantages, they requires a reorder point set for every item, that means in its turn that some type of forecast must be set for every item, often there are thousands of items to treat. (If the information in the bills of materials is used only forecasts for top-items are necessary.) Another great disadvantage is the lack of forward visibility, no calculations of forward machine loads can be carried out, and no automatic forecast is generated to the suppliers. The connections between items in the bill of materials and a Master Production Schedule (MPS) are not used; therefore no increases or decreases in production can be planned with ordinary reorder point systems. For CTP on the contrary, expected demand can be forecasted (increasing, decreasing or constant) for each semi-manufactured product that the end items consist of. For each semi-manufactured product the future demand of their underlying items can automatically be calculated. This means that the production of semi-manufactured products or modules can start in advance but the final assembly or production starts when actual customer orders arrive. To make semi-manufactured goods based on forecasts and finished products on customer orders is not possible using MRP; it must be possible? Why can not that be done using MRP? If you are not very well initiated in this problem it might seem a little bit strange. The answer is that to prevent old forecasts and new customer orders from ending up in a large mess support systems (such as those mentioned earlier) are needed when using MRP. But, not when using CTP, because forecasts and MPS are incorporated in MRP in a totally different way than forecasted demand is in CTP. CTP, unlike regular order point systems, makes it easy to present forecasts to suppliers. MRP usually gives you numerous order quantities and delivery dates in the future that never actually will occur with the predicted quantities and dates. With Cover-Time Planning to the suppliers are presented expected demand rates and for what periods they will last. 3 CTP presents an inherent priority CTP is not only advantageous to use when dealing with customer order driven production. It also gives an automatic basis to prioritize different manufacture and purchase orders. A short cover-time always means that that particular item is urgent, but it is hard to compare with other items. If you from current cover-time for one item subtract the lead-time and get a negative result; then you are about to run short and it is extremely urgent. If a machine is used to manufacture several different items they should be prioritized, the different items with their replenishing jobs should be manufactured in order after cover-time minus lead-time. The job with the smallest, or negative, number first and then in rising sequence. It is a way of providing priority to avoid future shortages. Observe that if you have long supply chains and several open orders for replenishment of the same item, it can be more urgent with the first order than what the cover-time tells. Also with CTP when you start/register a new production order or purchase order it is important to register a planned finish date and start date for that order. And between these two dates distribute start dates for the processes situated in between in different machines or work centres. Such as these dates, together with existing (current) cover-time and lead-time, can work as guidance for the production. 4 Control Work-In-Process and establish a cyclic policy It is common that companies register planned orders suggested by MRP a long time ahead, two weeks or more, i.e. change them to open orders. They don t just register them; they also release them, not only current week s orders but the orders for next week and the week after that. They start production and purchase because they think: Why don t do it now if it nevertheless must be done later. This is not a correct way to deal with it. Even if someone like to have a queue of fake sharp orders to make the purchase department to work and buy supplies that they believe are missing. This implies that a lot of material is put into work-in-process (WIP). Production orders usually vary in actual lead-time sometimes they go faster and sometimes slower than average. Sometimes a later manufacturing order even catches up with an earlier one (for the same item). The operator
then merge the two orders to save some setup time in the machine, but they forget that another order can get delayed if this is done. Too much WIP leads to longer total lead-times (cf. [2], [3]). It turns into a jumble and when the end product is meant to be assembled there are maybe three batches of several components in the inventory but one component is missing, so the end product can not be put together and delivered on time. Whether you use CTP or MRP too much WIP always is damaging; Curt Nicolin [4] meant and showed that the release of too much WIP creates future delays in delivery times; therefore you should not release or start production at a higher rate than what you complete (cf. [5]). In a mixed series production where production of different details and items recurs with more or less uniform intervals, then it is preferable to establish a cyclic policy. A cyclic policy that is established already when the orders are registered and not only in one single machine to make sure that the chosen sequence will reduce the total set-up time just in actual machine. If the setup times are highly sequence dependent it should be made sure that the production of the items occurs in a suitable order already when released. When the orders are started they should be started in a sequence that gives the shortest set-up time; e. g. if the following cycle has been decided: A1, A2, A3, A4, B1, B2,, B17, C1, C2, ; which also can count for items that not have a sequence dependent set-up time, the cycle is meant to create an even flow and balance between different items. If it is time to start manufacture A2, B16, B2, A4 because their cover-times are too short, they will be released in the following order: A2, A4, B2, B16 and all operators also know that they should be manufactured in the same order as the master cycle state above. This is to prevent clusters of the same item and not too much of one item and too little of another at the final assembly. 5 Work-In-Process is restricted with CTP Using CTP if there are no withdrawals of the end items there will not be any withdrawals of sub-assemblies and in turn no withdrawals of their components as in a Kanban or CONWIP system. CTP tries to hold the total amount of inventory and WIP above a limit, but CTP also restricts the total amount to another limit in the following way: n Q Supply ( n + 1) Q (5) Where n is an integer, Q is the used order quantity for the item in question and Supply according to eq. (1). This means that on hand inventory plus planned replenishments varies between these two values. But CTP does not (as Kanban) control WIP at every work centre. Spearman et al [6] point to that the number of Kanban cards has to be calculated for every single process in a way that prevents bottleneck starvation. If the bottleneck moves due to a change in the production mix the number of cards must be revised. Spearman and Zazanis [7] write that the effectiveness of pull systems does not result from pure pulling as in Kanban but from limiting WIP. CTP controls the total of what is in stock and upstream replenishing for every particular item the lead-time, buffer time and the demand rate in use; which is more like CONWIP (cf. [6], [3]). Using MRP the planning horizon must be at least as long as the cumulative lead-time of the product to prevent rolling horizon disturbances. If the planning horizon is too short, the new demand in the end of the planning horizon in the current MRP-calculation, will create a need for purchase of items on the lowest structural level (in the bill of material) in bygone time, With MRP all demand on upstream level does not disappear completely until the available inventory status is positive during the entire planning horizon for the item. MRP does not immediately consider stops or delays in production. Orders are released to follow planned orders set by the MPS, so even if there is no use for item A because item B is missing for assembly, production for item A will still continue. This leads to a situation where one process is working at full capacity even if the production is not in balance. Too few end items may be manufactured but the inventory of subassemblies and components increases. Therefore controlling WIP with MRP and with a long planning horizon the system reacts slower than CTP. Comparative simulation studies ([8], [9]) also show that MRP create a larger variation in WIP than CTP when the coefficients of stochastic variability for demand and lead-times are increased even for small examples with only 10-12 items (in the bill of material).
6 Just-in-Time production Just-in-Time production requires accurate information. When using MRP this requires a lot of calculations, which are complex and therefore time consuming. The number and complexity of the transactions in MRP may limit the frequency of recalculations. Every time a calculation is made it should also close the loop (MRP calculation, Work load analyses, Adjustment of MPS, MRP calculation etc.) When CTP is used no special preparation is needed before the calculation and the calculation itself is simple. With Cover-Time Planning new demand rates are only necessary to be recalculated when the forecasts for the top items used in the forecast calculations have changed. The main calculation of CTP (test of eq. (3) and (4)) can easily be done a couple of times a day and as a result it gives more updated information for Just-in-Time production. 7 Estimated work load and CTP The suggested planned orders which are generated by MRP are used together with already open orders and registered setup and production times for every item to calculate the work load in different work centres (Capacity Requirements Planning). The work load calculation gives the user a useful hint of the future demand of capacity, which an ordinary reorder point system does not provide. For CTP the current demand rate determines how many working days already released orders will last, and then a new order must replenish inventory. The prevailing demand rate decides how many working days this order quantity will last etc. until the end of the calendar used. These calculated proposals together with already released orders can then be used in the same way as for MRP calculate the work load in different work centres. The fact that replenishment is also a withdrawal on the underlying structure level in the bill of material is not considered in these calculations. The error in time phasing is in most cases negligible from a practical point of view, the important thing being that the work load is correct according to the volume of setup and production times. (The planned orders of MRP with its time-points and quantities will never either in practical cases be realized exact according to plan.) By applying this procedure, the same types of work-load reports as with MRP can also be presented by CTP. 8 Importance of, and principles of CTP A most important process for the company and its supply chain is the way it determines and promises delivery times and quantities for incoming customer orders. The technique (MRP or CTP) used and the company s ability to use it will decide the level of customer service, how much capital will be invested in work-in-process and inventories, what efficiency and costs will be created in the production facility, and in the end what financial result the company achieves. It is not financial targets that decide an organisation s long-term profitability it is the way it organises its work (cf. [10]). CTP is not a completely new idea as can be seen from the references. From the beginning the interest from software companies was half-hearted. (Interesting technique, but it is unproven! Why disturb already made investments in expensive software?) But, quite recently the interest for CTP has increased also from software companies. IBS has incorporated the technique in new releases of its ERP-system ASW (http://www.ibs.net/), and Monitor in its ERP-system (http://www.monitor.se/). Hopefully this in the future can result in practical cases from real installations. The greatest advantages with CTP compared to MRP are in customer driven production with many variants of different product families, and with products with several levels in the product structure (bill of materials). Deep structures is definitely not recommended, but sometimes it is inevitable; and then CTP avoids better than MRP the bullwhip effect, because the demand for underlying levels are undistorted of order quantities. The main principle behind the idea of CTP is: wait as long as possible before production starts; wait so that it is known exactly what the customer asks for. By that material will not be tied up in something that the customer is not asking for, and result in material shortages for another product that the customer really is asking for. Then basically no unnecessary production is made, and the production is kept lean. When the cover-time is long
compared with the lead-time, one can wait and put machines on hold for maintenance, and/or move man power to another machine. 9 References [1] Segerstedt, A. 2006. Master Production Scheduling and a Comparison of Material Requirements Planning and Cover-Time Planning. International Journal of Production Research, 44 (18-19), 3585-3606 [2] Silver, E. A., Pyke, D. F., Peterson, R. 1998. Inventory Management and Production Planning and Scheduling (3 ed.) (John Wiley & Sons) [3] Hopp, W. J., Sperman, M. L. 2000. Factory Physics (2 ed.) (McGraw-Hill) [4] Nicolin, C. 1959. Planering av en verkstad med blandad tillverkning matematiska tumregler till hjälp för att organisera effektiv verkstadsplanering (Production control in a workshop with mixed production mathematical rules of thumb for organizing an efficient production: in Swedish). Arbetsrapport PM VD4a/59, delaval Jungström, Finspong [5] Segerstedt, A. 1999. Escape from the unnecessary - some guidelines for production management, Production Planning & Control, 10 (2), 194-199 [6] Spearman, M.L., Woodruff, D.L., Hopp, W. J 1989. A Hierarchical Control Architecture for Constant Workin-Process (CONWIP) Production Systems. Journal of Manufacturing and Operations Management, 2 (3), 141-171 [7] Spearman, M.L., Zazanis, M.A. 1992. Push and Pull Production Systems: Issues and Comparisons. Operations Research, 40 (3), 521-532 [8] Segerstedt, A. 1991. Cover-Time Planning - An Alternative to MRP, (Linköping: PROFIL 10) [9] Segerstedt, A. 1995. Multi-Level Production and Inventory Control Problems - Related to MRP and Cover- Time Planning. (Linköping: PROFIL 13) [10] Johnson, H. T., Bröms, A. 2000. Profit beyond measure. (The Free Press)