Is just-in-time applicable in paper industry logistics?
Juha-Matti Lehtonen, Jan Holmstrom. Supply Chain Management. Bradford: 1998. Vol. 3, Iss. 1;
The applications of just-in-time (JIT) logistics specific to the paper industry are scarce and
somewhat conflicting. The potential contribution of alternative logistics systems in the paper industry, and the scope for efficiency gains through the application of JIT logistics systems are
? Jump to indexing (document details)
Full Text (6824 words)
Copyright MCB UP Limited (MCB) 1998
Juha-Matti Lehtonen: TAI-Research Centre, Helsinki University of Technology, Helsinki, Finland and Jan Holmstrom is based at the Department of Industrial Management, Helsinki University of Technology, Helsinki, Finland
Jan Holmstrom: Department of Industrial Management, Helsinki University of Technology, Helsinki, Finland
This paper is based on the results of a national research project on the logistics situation of the Finnish fine paper industry. The framework of the research project was to analyse the present situation thoroughly, suggest new logistics solutions for improved performance in outbound logistics and production control and to develop a supply chain model to test the effects of the new logistics solutions. The project was carried out by the IIA Institute at Helsinki University of Technology and financed by the Technology Development Centre of Finland (TEKES).
The Scandinavian forest industry has gradually developed from being essentially a manufacturer of bulk products like pulp and newsprint towards more diversified andcustomer specific printing and writing grades. At the same time the logistics solutions of the industry have not adequately reflected the change that has taken place in its products. For example, the finished goods inventories of the Finnish paper industry are considerably larger than those of major competitors(Lehtonen, 1995). The primary objective of the paper is to identify the critical factors for successful logistics in the paper industry. This general objective can be further focused on twospecific research questions: - Where should order penetration-point (Sharman, 1992) be located in the paper industry? - Is just-in-time (JIT) applicable in the paper industry logistics?
The research method is multiple case study. The cases were selected so that they included both Nordic and Central European paper mills as well as producers of fine paper,speciality paper and bulk grades like newsprint and SC (supercalendered paper). All the cases were executed in co-operation with a large consultancy company. This imposed restrictions on the case selection process but provided access to case companies and information that would not have been forthcoming without their involvement.
The scenario simulation methodology has been used for testing the effects of logistics alternatives in the paper industry. It is based on both existing theory on simulation projects and the Controllability Engineering methodology (Eloranta and Raisanen, 1984). The scenario simulation methodology means that the present logistics operations of a case paper mill are first modelled as a reference point and then all the changes are simultaneously implemented in a, hopefully, improved scenario model. The improvement potential can then be assessed as the difference between the base case and scenario model results. The simulation studies were carried out with a commercially available simulation package (Taylor II by F&H Simulations BV).
In order to assess the improvement potential, performance measures have to be used. Stalk and Hout (1990) argue that the single most important performance measurement is time. For logistics system, the main performance measures are availability, capability(= delivery time and reliability) and quality (Bowersox et al., 1986). In this study the performance of supply chain operations is measured by lead time and amount of inventory. In the simulation models where there was either make-to-stock or assemble-to-order production control, the amount of inventory was used as a performance measure while either fill-rate (cases 1, 2 and 3) or availability (case 4) was held constant. In make-to-order production (cases 1 and 3) the main performance measure applied was lead time. Reliability was not explicitlystudied because it is also a component of lead time. This is especially true in simulations using a first-in-first-out scheduling rule and a paper machine cycle. Just-in-time logistics and choice of order penetration point
There are many views on the logistics aspects of JIT. Of the key features mentioned by Sakakibara et al. (1990), the ones directly connected to logistics are JIT delivery from suppliers, small lot sizes and, to some extent, a repetitive master production schedule. Harmon (1992) advocates JIT small lot production with cellular manufacturing, quick changeovers and pull scheduling. According to Crum and Holcomb (1994) the adoption of JIT production and inventory systems increases the need for smaller, more frequent shipments and quick response times.
According to Bowersox et al. (1986) a logistical performance cycle consists of nodes and links. A node is a physical location, for example a factory, a warehouse or a customer. Transportation and communication connections are referred to as links. Bowersox et al. (1986) claim that the essential interfaces and decision processes of a logistical system can always be described in terms of individual performance cycles. In other words, they claim that any logistical system consists of individual order-delivery processes. However, their concept of logistical performance cycle does not show or take explicit account of the decision rules (e.g. whether a replenishment order is sent or transport arranged according to a schedule, up to some consolidation limit or on demand).
If two logistical performance cycles are arranged sequentially into line, we arrive at a logistical or supply/delivery chain concept. Figure 1 shows one view of a supply chain. The supply chain extends from raw materials procured by supplier, through factory production including master production schedule (MPS) orders and distribution to the end customers (Jones and Riley, 1992). Other authors have also made popular the idea of a logistical chain. According to Schonberger (1990) a supply chain can be viewed as analogous to a factory, when only the machines in a factory are replaced by factories in the supply chain. Porter (1985) has defined the widespread value chain concept. Jahnukainen et al. (1995), Oliver and Webber (1992), and even Bowersox et al. (1986) discuss the concept of a logistical chain. The advantage of a logistical chain concept is that it takes a broader view of the logistical process by overcoming the physical borders of an individual enterprise.
There exists a well-known production control choice classification of make-to-order (MTO), assemble-to-order (ATO) and make-to-stock (MTS) (Vollman et al. 1992). Make-to-stock means that a customer order is filled from the finished goods inventory, assemble-to-order that the customer order is assembled from the sub-assemblies inventory and make-to-order means that the customer order is fully manufactured to the order from the raw materials either on hand or to be ordered.
Lehtonen et al. (1996) have furtherelaborated the MTO/ATO/MTS classification especially in implosive industries where a range of products are manufactured from a few common raw materials (Burbidge, 1994). According to Lehtonen et al. (1996) the limiting factor for production control choice in the implosive industries is the length of customer acceptable delivery time compared to production planning and manufacturing cycle time plus the transport time to the customer. In make-to-order the production can be directly controlled by customer orders. No finished goods inventory is needed. In make-to-stock the products have to be delivered from the finished goods inventory to meet the delivery time requirements. Production is controlled by more or less accurate demand forecasts. Delivering a standard range of products from stock is not necessarily a bad idea since it provides fast customer service. In the implosive industries, however, the large number of end products enforce sizeable inventory investment, thus making delivery from finished goods inventory an undesirable choice.
When it is not feasible to plan the production flow by customer orders, there could be a point in the production flow from which the full product range can be planned,manufactured and transported to the customers within the delivery time. This point in the production flow can be used as a control point. Before the control point the production is controlled by consumption and thereafter by orders. Products are manufactured and delivered according to customer ordersfrom a semi-finished inventory which is held at this point. As the product variants in the implosive industries increase in each processing phase, the control point should be set as early as possible in the production process.
Sharman (1992) defines the concept of order penetration point as being a point in the logistical chain at which a product is assigned to the specific, known customer. Before that point, the flow of materials is controlled by forecasts and plans and afterwards by customer orders. The order
penetration point is the last point where inventory is held. It is also in most cases the point where product specifications are frozen. Each product design has its optimum order penetration point that depends on the balance of competitive pressures in the chain, product cost, product complexity and target customer segment requirements.
The contribution of the order penetration point concept over MTO/ATO/MTS classification is on the make-to-stock side. The manufacturing view of make-to-stock order filling from finished goods inventory is widened to include more possible order penetration points along the distribution of the product, thus making the order penetration point concept more logistically oriented than its manufacturing counterpart. The weakness of the concept in a logistical chain is the vague definition of a "customer" because in a chain each echelon is a customer of the next.
According to Mertens and Houben (1996) there are four different possible orderpenetration points, which they call decoupling stocks, in the paper production process. These choices are shown in Figure 2. The order penetration points and their advantages are:
(1) Decoupling stock of base materials. Make-to-order production with long and difficult to control delivery times due to PM cycle and finishing lead time. Low amount of stock.
(2) Decoupling stock of jumbo reels. From PM onwards production to order. Drawbacks are difficult storage due to large jumbo reel size and slow quality feedback. Also due to the large size of a jumbo reel, small orders are hard to make both fast and economically.
(3) Decoupling stock of winder reels. Fast service can be provided up to two days or less if there is enough sheet capacity. It is also still possible to provide a high degree of customisation, including cut-to-order sheet sizes and a variety of different wrappings. The drawback, in addition to the stock itself, is the possible increase in winder and sheet trim loss.
(4) Decoupling stock of sheets on pallets (finished product). It can be located either at the mill or at distribution centres. The advantage is a very short delivery time. The disadvantage is that many different sheet sizes, ream heights and wrapping papers make the required inventory investment large.
According to Harmon (1992) one requirement to achieve JIT is machine over-capacity. In a multi-product and multi-machine production system changes in product mix cause machine loads to vary. As Huq and Bernardo (1995) point out, even with adequate system capacity, bottleneck areas surface due to shop load fluctuations which significantly reduces the effectiveness of a pull system. BothHarmon (1992) and Huq and Bernardo (1995) refer to a multi-machine job or flow shop. Lehtonen et al. (1996) recognise in their quick response system vision for implosive industries the need for over-capacity in order to cope with short-term demand fluctuations, which in effect amounts to a chase capacity strategy.
Most cases of process industries, including paper, are notable exceptions to the over-capacity requirement. While the machines are bought from manufacturers supplying everyone in the industry, the competition is usually fierce and equipment utilisation can make the difference between profit
and loss. The competitive conditions force companies to operate at near-full capacity (Harmon, 1992).
A somewhat different view is taken by Eloranta et al. (1994). They claim that the most obvious example of misunderstanding the virtue of speed is found in the Finnish forest industries. Inventories rose from 45 days in 1979 to 75 days by the end of the 1980s (Holmstrom, 1994). Eloranta et al. (1994) argue that this development could only be explained by technocratic pattern of thought, where the industry continues to invest in ever faster and larger machines while the markets want more diversified products in smaller and smaller deliveries.
Table I presents a summary of the cases, including situational factors, performance measures, improvement actions and results that are the improvement potential assessments. In the following sections the cases are discussed in detail. Each section starts with a short case description. The base case is described and base case simulation results are presented. The re-engineering actions that are applied in the scenario model are then described and the and the simulation results presented.
Case no. 1
The case mill is located in Central Europe and has a predominantly domestic customer base. The main products are three different qualities of speciality papers. For some grades additives are sometimes used. Even as a speciality mill, its annual capacity is relatively small. Some 30 per cent of the production is sheet while the rest is delivered in reels. All the deliveries are direct deliveries to the customer who are either converters or wholesalers.
Figure 3 shows the delivery chain of the case 1 mill. The mill has a finished goods sheet inventory inside the plant area for two qualities of main weight and size combinations from which fast deliveries can be made. This policy is shown in Figure 3 with a small rectangle representing the orderpenetration point (B). Unstocked sheetquality, weight and size combinations, and all the reel orders are made to order at the paper machine. In Figure 3 this policy is shown as order penetration point A. The paper machine cycle is two weeks, which meansthat every quality, weight and additivecombination can be produced only once every fortnight.
The data used for the in-depth analysis were obtained from the company's database. The sample was three months' deliveries. Additional information was provided by company documents and interviews. Delivery time analysis was performed by calculating the time between order date and delivery date. The analysis showed that a large number of sheet orders are delivered from stock inside one week from order placement. However, closer inspection revealed that these orders are of relatively small volume. The delivery time for reel orders, which are made to order, ranges from three to five weeks.
The controllability engineering analysis identified a long delivery time for make-to-order reel deliveries. Also it was found that the finished goods sheet inventory was large. As a result, the problem statements for the simulation study were to:
- determine the impact of paper machine cycle and intermediate reel storage on customer lead time and delivery performance for sheet deliveries
- determine the effect of setting up intermediate reel storage on finished goods sheet inventory. Base case model
The scope of the mill's present operations model was defined to include order book, paper machine, sheeting, wrapping, stocking and delivering operations. The model input data included demand, paper machine cycle length, machine capacities and inventory control logic. The performance measures were inventory and delivery time. The paper machine cycle length used was two weeks. The demand was modelled based on approximations of the actual mill demand that were obtained during the controllability analysis phase.
The base case simulation gave the result for inventory as 950 tonnes of which 400 tonnes was in sheets and 450 tonnes in reels, with an average lead-time of 14 days.
In the scenario model two important changes to the present operating model were introduced: (1) The paper machine cycle length was halved to one week.
(2) An intermediate reel storage was introduced after the paper machine but before sheeting. The scenario delivery chain is shown inFigure 4. Sheeting is carried out to order from the intermediate reel inventory. This enables fast sheet deliveries from the reel storage and replaces filling the order from paper machine for all but the largest orders. Intermediate reel storage also enables eliminating standard sheet stock without compromising the delivery time for sheets which are at present made to stock and delivered from the standard sheet stock.
The scenario inventory was 1,050 tonnes of which 660 was intermediate reel storage, 130 sheets and 260 reels. The simulation results for delivery time are shown in Figure 5. The customer lead time distribution was reduced substantially and the average delivery time was halved. This also reduced the cycle inventory for reels. The sheet deliveries could be made in just one day from the intermediate reel storage. In further simulations it was noticed that the delivery time of sheets was now very much dependent on sheeting capacity utilisation. For the presently stocked sheet qualities this meant a reduction in inventory levels with no negative effect on the customer lead time! Case no. 2
Case no. 2 is a relatively small paper mill with a yearly production of speciality papers under 50,000 tons. The mill is located in Western Europe. The study is based on production and delivery data over a six-month period. During that time 18 different grades were produced. The speciality of the mill is various colours. Coloured paper manufacturing imposes a rigid cycle because pulp itself has a yellowish hue that forces the colours to be produced in a cycle starting from light blue and white. Each colour change imposes a set-up and settling time. Production of various grades and colours makes the production environment difficult and the production cycle time was six weeks at the time of the study. The majority of the production (86 per cent) was sheet and the rest was sold in reels. The deliveries were direct deliveries to the customers who were wholesalers or sales agents. Figure 6 shows the delivery chain of the mill. The case 2 mill had a large finished goods sheet inventory from which fast deliveries could be made. Make-to-stock accounted for 70 per cent of the production volume. In Figure 6 this is shown as (B). The remaining 30 per cent was evenly divided between reel and sheet production to customer orders, shown as (A) in Figure 6. An analysis of operational performance was carried out based on the sample data with additional information from company documents and interviews. The top level analyses that were performed were delivery time, reliability and demand.
Inventory was derived from the company documents. The finished goods sheet inventory was equivalent to two-and-a-half months of demand. The intermediate reel storage size was equivalent to demand for one month.
The delivery time analysis was performed by calculating the difference between order date and due date. About 80 per cent of the volume was delivered within a week of the order date. The demand analysis proved that the total demand of all grades was very stable. The main constraints for the production turned out to be the very small order sizes in combination with a wide product range. Almost 80 per cent of the cumulative sum of orders consisted of less than half a tonne (i.e. less than the size of one pallet).
The main problem area in the case 2 mill was production and inventory control. Unlike case 1, sheeting-to-order was deemed unfeasible from the start. The minimum feasible sheeting batch size was one reel which equals 1.5 tonnes. With this constraint and the existing order size distribution, sheeting to individual orders is not feasible.
Intermediate reel inventory existed already. Based on the controllability engineering analysis, the problem statement for thesimulation study was simply:
- how to decrease the finished goods inventory?
Base case model
The present operations model consists of paper machine, intermediate reel inventory, sheeting machine and finished goods sheet inventory. The model input data include sheet demand for each sheet size and paper machine cycle. The customer demand is filled from the sheet stocks. Sheet stocks, in turn, are filled from intermediate reel inventory. The sheet stock level control is based on reorder point logic. Intermediate reel inventory is replaced from the paper machine. The intermediate reel inventory control logic is periodic lot-for-lot replenishment so that the sum of intermediate reels and sheet stocks are replenished at equal level after each period. The simulation results in Figure 7 show the amount of inventory needed in the base case model to achieve different order fill rate levels for the main grade. Each dot is the average of three simulation experiments where each experiment consisted of 67 cycles. The average inventory requirement increases progressively as fill rate approaches 100 per cent. Specifically, 96 per cent fill rate is achieved with approximately 18.5 tonnes average inventory which requires 32.5 tonnes of initial stock in the modelled periodic review system. This can be compared to the 30.3 tonnes of expected demand for the main grade during the cycle.
Two scenarios were tested with the case 2 simulation model:
(1) reduction in the number of sheet sizes
(2) reduction, by half, in the paper machine cycle.
The reduction in the number of different sheet sizes was achieved by deleting the two lowest selling sizes and allocating their demand to the remaining ones based on their present demand proportions. The first simulation experiment was done with the same initial values as in the present model. The results were that reduced sheet number scenario had 0.1 tons lower inventory and achieved 97 per cent, i.e. 1 per cent better fill rate.
Halving the paper machine cycle length from the present 45 days to 23 days had a clear effect on the inventory. With present situation initial values, a 99.6 per cent fill rate with 25 tonnes average inventory was achieved. Next, the initial inventory was reduced to 20 tonnes or 38 per cent. This time the fill rate was 97 per cent and the average inventory 13.1 tonnes. Even though the fill rate was increased by 1 per cent the average inventory was reduced by 5.4 or 29 per cent. The experiment was terminated when it was evident that reduction in the paper machine cycle could result in a sizeable inventory reduction for the main grade of case two mill.
In conclusion, the change actions to reduce inventory that were tested were halving the paper machine cycle and reducing thenumber of different sheet size alternatives. Reducing the number of sheet sizes from five to three decreased the inventory only about 5 per cent at a 95 per cent fill rate. Halving the machine cycle had a strong effect on inventory, cutting the average inventory by 29 per cent even at a slightly higher 96 per cent fill rate.
Case no. 3
Case no. 3 is a large Nordic fine paper mill. The aim of the study was to create a scenario for the future logistics of the mill. Thisscenario could then be used by two major logistics service providers to give an idea of the future logistics choices of the case 3 mill and how these affect its logistics service requirements. The mill itself was not involved in the study and a controllability engineering-analysis was not performed. Instead of this, a major consultancy provided the necessary input data.
The main market of the case 3 mill is Western Europe. The case mill has two paper machines, one of which manufactures six and the other eight different grade and weight combinations. Sheet orders amount to 65 per cent of the production and the rest are reel orders filled from the paper machine. At the base case, the mill has an intermediate reel inventory from where small and medium sized sheet orders can be filled. This is shown as (B) in Figure 8. All reel orders and large sheet orders are filled from the paper machine, shown as (A) in Figure 8. However, due to infrequent and long transport times, the delivery time to the customer from the intermediate reel inventory is not fast enough in most cases. Long lead time in combination with the sheet wrapping at the mill forces the mill to hold a customer dedicated inventory in port terminal near the customer for all orders. Deliveries can be called off very fast fromthis port terminal. This is shown as (C) in Figure 8.
The problem statements for the simulation study were to:
- test the effects of an alternative transport arrangement and shorter paper machine cycle - assess the intermediate reel inventory size at 98 per cent fill rate.
The model included the paper machines, intermediate reel inventory, sheeting despatching, transportation and terminal inventory at the market area. However, the actual customer call-off in the make-to-stock production, shown as (C) in Figure 8, was excluded from the model. The utilisation level of the mill was held constant so that the order book could be settled at around four weeks which was also the production cycle length.
The model included two markets. Market 1 had a shipping time of 60 hours and market 2 had a shipping time of 84 hours. Port operations length were one day on both domestic and foreign port. There were weekly sailings to both markets. The consumption was evenly split between the markets.
Base case model
The lead time from order to the arrival of paper at the foreign port terminal is shown in Figure 9. The lead time median for the small and medium sized sheet orders was nine days and the maximum was 12 days excluding the small tail. The median for the reel and large sheet orders which are made-to-order was 34 days.
As can be seen from Figure 9, no orders had a lead time less than six days. This is due to the shipping and port handling times. The lead time distribution in these deliveries was due to the small transport time difference between the two markets and the low shipping frequency. There is a small tail in thedistribution which is attributable to an out of stock situation in the intermediate reel inventory. The orders delivered from the paper machine show a wide distribution that was caused by the four-week-long paper machine cycle.
The simulation results for the base case average total inventory was 33,900 tonnes and without port terminal inventory 22,000 tonnes. With the four-week-production cycle, the intermediate reel inventory needed to reach 98 per cent fill rate was 10,000 tonnes.
The scenario was similar to base case except for two changes:
(1) an alternative transport arrangement
(2) two-week-production cycle in the paper machine.
The alternative transport arrangementmodelled a concrete, real alternative where a daily train was used to transport the paper to a new domestic port with three sailings per week to both markets. The total transport time remained the same for market 1 but decreased by four hours to market 2. Scenario lead time simulation results are shown in Figure 10. The lead time from order to arrival at the foreign port terminal for small and medium sized sheet orders was reduced considerably. The median was only 4.5 days and the maximum lead time dropped from two weeks to one, as a result of the alternative transport arrangement. The median lead time for the large sheet orders and reel orders that are made-to-order fell to 18 days, primarily due to the shorter production cycle. Only a few days are caused by the alternative shipping arrangement.
In the scenario all inventories, except for the in-transit inventory were reduced. The average total inventory was 14,400 tonnes and 13,100 tonnes without the foreign portterminal inventory. The finished goods inventory at the mill is now negligible because each day the train takes the production to the domestic port. The terminal inventory is reduced because replenishments three times a week match the consumption more closely than the once-a-week replenishment. The intermediate reel inventory needed to achieve 98 per cent delivery is now only 5,200 tonnes. The reduction in the intermediate reelinventory requirements is attributable to the shorter production cycle.
Case no. 4
Case no. 4 is a large Nordic newsprint mill with several newsprint machines producing standard newsprint grades. The case 4 mill has both foreign and domestic customers but its main markets are in Western Europe. The mill has a customer dedicated call-off stock at the port terminal from which deliveries can be made by customer call-off. This port terminal stock is replenished from the