Discuss the essential differences between the traditional factory and the new cellular manufacturing approach at Applicant
Applicon Case Study
A review of the Applicon Company’s position is attached.
1. Prepare an analysis of the company with particular reference to the following:
a) the characteristics of the market served by the company;
b) the length of lead time. Prepare an appropriate bill of materials for the company’s traditional factory, find the longest cumulative lead time for the process, and comment on the length of the lead time.
c) A university has ordered for fully tested 200 of the 4620 Alpha terminals for their CAE laboratory to be delivered within 4 months’ time. Assuming that the initial plan demanded for 40 finished terminals per week and, with the following starting inventory, determine the latest planned work order releases for the Table, Pen, Agid PCBs and Cover Assembly as particular items, and Cable as a general item. The default pre-testing safety stock is 10 unless specified.
The starting inventories (units) are:
Item Opening inventory Lot size Safety stock
4620 terminals (Pre-integration test) 50 10 10
Base Electronics Assembly 50 10 10
Monitor Assembly 40 10 10
Keyboard/tablet Assembly 60 10 10
Keyboard Assembly 70 40 10
PCB 60 20 40
Assume that the factory works 5 days a week and lead times are to be rounded to number of weeks for this section.
2. Discuss the essential differences between the traditional factory and the new cellular manufacturing approach at Applicant. Consider items such as material flows, process lead times, inventories, shop floor order quantities, quality issues, and the impact on supporting the market.
3. Discuss the specific changes that were made to the firm’s manufacturing planning and control system as a result of the investment in the new production process.
4. Build a Witness model of the traditional Alpha production system. Analyse the results and comment on how performances could be improved.
5. Suggest what applicant should do to bring about further new improvements.
6. Structure and presentation of report.
Any work extracted from texts must be clearly annotated as must all figures and diagrams. The work must be referenced, contain a bibliography and be divided into suitable sections.
The typed report should be around 2,500 words (excluding tables) and must be submitted by ………………….
Applicon was founded by a group of venture capitalists and MIT professors. The company, which is now owned by Schlumberger, designs and manufactures computer aided engineering (CAE), computer aided design (CAD), and computer aided manufacturing (CAM) systems. In 1997 Applicon had a total of 1,000 employees and sales of approximately £100 million. All manufactured products are produced in a single facility located in Billerica, Massachusetts.
The high-end CAE products include sophisticated systems used for various analytical engineering applications. The software for these high-end systems is proprietary and runs on DEC and Sun central processing units with Applicon designed workstations. The low¬-end CAM systems use Applicon software, Sun and Tektronics workstations, and DEC VA processors. Major applications of such CAM systems include robotics and numerical control machines.
Applicon’s first product line, the Applicon Graphics System (AGS), was introduced in 1984. This system was successfully sold in the electronics market during the 1980s, with sales peaking in 1990. Although the AGS was a profitable product line for Applicon, it was felt in late 1990s that the electronics market at which it was aimed had become saturated. The real growth was now in the mechanical market. In this market, CAD/CAM systems were increasingly being applied by such users as the automotive industry and machine shops.
These market factors were instrumental in Applicon’s decision to introduce the Series 4000, an entirely new product line targeted at the mechanical market. The Series 4000 was announced in the spring of 1991 with the first shipment made in May 1992.
The introduction of the Series 4000 was a major turning point for the company. Applicon had entered an extremely competitive market with a new and unproven product. Competing against such CAD/CAM system producers as Computervision, Intergraph and Autodesk, Applicon faced tremendous start-up problems. One result of these conditions was a serious shortfall in actual sales as compared with forecasts. At the time, however, management did not react to the shortfall in demand. The belief was that sales would eventually increase once the introductory ‘bugs’ had been shaken out. Thus no action was taken and Applicon continued to build to finished inventory in anticipation of increased demand.
Another problem was the higher than expected cost of introducing and manufacturing the Series 4000. The importance of reducing product costs was quickly becoming apparent. Unlike the electronics market, the price-to-performance ratio was a critical issue to the price sensitive CAD/CAM consumers in the mechanical market.
Large product start-up costs were incurred. These included training costs of field engineers, applications engineers, and the salesforce. Frequent engineering change orders and reliability problems also contributed to the high costs of product start-up by making obsolete large amounts of finished inventories. Overhead costs were another contributing factor. Increasing costs of MIS, materials management, production management, inventory control and quality control were the norm.
Applicon was also losing its competitive edge because of an inability to respond quickly enough to changes in technology. It was not uncommon for radical hardware and software advancements to occur on an almost monthly basis in this dynamic market. Applicon had to become much more responsive to the frequent design changes characterizing the industry.
Lastly, Applicon was now dealing with a more sophisticated customer base than in previous years. These consumers had a better understanding of both the hardware and software of CAD/CAM systems. They not only developed higher expectations for price-¬to-performance ratios and delivery responsiveness, but also demanded a higher quality product. Applicon’s quality improvements of one or two percentage points per year were not sufficient, and something needed to be done to improve quality dramatically.
The traditional factory
Under the ‘traditional’ mode of operation, production would build to stock in accordance to an annual build plan, as estimated by marketing. The perception was, ‘if the finished items could not be sold it was a sales problem, not a manufacturing problem’.
Using the annual build plan, a master production schedule (MPS) was created for each month’s production of end-items. These MPS requirements were then input into a materials requirement planning (MRP) system. The MRP system generated the work orders for the lower-level items by exploding through the bill of material (BOM), applying the ‘gross-to-¬net’ logic, and off setting for lead times. These work orders were used to schedule and control production on the floor.
Figure1 is a simplified indented bill of materials for a typical product, the 4620 Alpha terminal. This BOM shows the in-house manufactured items that support the top-level product. The BOM also identifies the four types of manufacturing operations involved in producing the 4620; top-level assembly, lower-level mechanical assembly, cable assembly, and printed circuit board (PCB) production. Figure 2 shows the complete terminal and the three top-level assemblies that make up the terminal.
The detailed production steps and material flows for each of the manufacturing operations are shown in Figure 3. Here, the production process was initiated with the release of work orders by the MRP system. These work orders were lot sized into one month batches. Material for one month’s worth of demand was kitted in the central stockroom for each of the subassembly (PCB, mechanical and cable) operations. These kits were then released to their corresponding production areas. Once completed, the batches, or kits, of finished subassemblies were sent back to the stockroom where they were rekitted for final assembly in the top-level assembly area.
The finished top-level assemblies were then integrated and tested as a single unit in the top-level integration/test area.
Once tested, the terminals went to finished goods inventory. Upon receipt of an actual customer order, the terminals and various purchased peripherals (plotters, tape drives, etc.) were released to system integration and test. Here, the complete system was tested, and ‘burned in’. After completing any necessary final touch-ups, the software and documentation were consolidated with the system. The complete order then underwent final inspection before it was shipped to the customer.
The various lead times associated with each of the processes are shown in Table 1. Figure 4 portrays the plant layout under this production system. Under this layout, the production operations (top-level) assembly, mechanical assembly, cable assembly and PCB production) and the integration activities are arranged in a functional manner.
This traditional MRP based production system was prone to a number of problems. These problems became serious when Applicon developed the Series 4000. Running the production floor to an annual build plan and ignoring actual customer orders resulted in large amounts of over-built finished inventories. By 1994, £5 million worth of products sat in finished goods (approximately four to five months’ worth of demand). In addition to the large finished inventory levels, £11 million of component inventories existed, and over twenty weeks of work in process was on the floor. These inventories not only tied up working capital, but also required extensive floor space. Furthermore, millions were lost as a result of stagnant inventories being made obsolete by the constant bombardment of engineering changes and new product introductions. Reserves of £100,000 to £115,000 per month were routinely set aside for such obsolescence.
Large overhead expenses were associated with the traditional production system. The numerous transactions in stockroom kitting, shopfloor control, and the production and financial control mechanisms were key drivers of these overhead expenses. Eighty-three employees were working in materials management and 60 in quality assurance (QA), as compared to only 160 actual line manufacturing operators.
The responsiveness of the traditional production system was too slow for Applicon to remain competitive in the dynamic CAD/CAM market-place. The four to five month manufacturing lead times were much too long in a market where major product design changes were occurring on a monthly basis.
In 1992 the plug-and-play rate was 85 per cent. The plug-and-play rate is the percentage of units sold to customers that contain no manufacturing defects and can therefore be plugged in directly and used. Improvements in quality were needed to quickly get this rate into the upper 90 per cent range. Quality problems also resulted in excessive rework requirements on the shopfloor. Under the traditional system, a lot, on average, had to be completely reworked one time through each of the four manufacturing operations (PCB production, mechanical assembly, cable assembly and top-level assembly).
Quality improvements were difficult to achieve under this system because work was scheduled in lots of one month’s worth of demand. Such scheduling in monthly batches caused the quality problems to be masked by inventories. Moreover, the low visibility of defects, combined with slow feedback to the source of the quality problems, did little to foster worker involvement in any quality improvement efforts.
Applicon came to realize that the manufacturing methods of the 1980s were now inappropriate for the highly competitive and dynamic mechanical market. In 1994, Don Fedderson, the company’s president at the time, issued a mandate to Tom Genova, the newly promoted vice-president of manufacturing: ‘Reduce costs and inventories, increase quality and responsiveness, and make Applicon competitive once again.’
Genova, being the ‘radical young engineer who was put into the factory by Fedderson to shake things up’, was given free rein to implement any changes deemed necessary. Genova concluded that just-in-time (JIT) production was the solution that Applicon was looking for. With the support and backing of Fedderson, Genova forced manufacturing to convert to JIT.
JIT production was first implemented on a single product model, the 4620 Alpha terminal, in May of 1994. Experience with the Alpha line then allowed for a full conversion effort, which started three months later.
This full scale conversion to JIT can be divided into three phases. Phase 1, which took place over the second half of 1994, focused on process conversion. Phase 2, which began in 1994 and lasted into the beginning of 1997, focused on resource management. Applicon is now in the third phase of implementation where major efforts are being directed towards improving vendor integration.
The conversion to JIT resulted in far-reaching changes at Applicon, many of which are still occurring. Production related changes include the layout of the shopfloor, the production processes, production scheduling and control, MRP and capacity planning, master production scheduling and order processing.
Layout and process improvements
One of the first actions taken when converting to JIT was to rearrange the equipment to achieve a smoother flow of production. In this respect the concept of work cells, in which equipment is grouped by product or similar product families, was introduced.
Concurrent with the relayout into work cells were efforts focused on process improvements. Assembly instructions were improved to help maintain the rapid flow of production characteristic of JIT. In addition, equipment and production processes were modified to reduce set-up times so that production could be scheduled in small, daily lot sizes.
The layout of today’s JIT based production facility is illustrated in Figure 5. The figure also shows the flow of materials through the plant. As can be seen by comparing Figures 4 and 5, the total area dedicated to manufacturing has been significantly reduced under the new layout.
Purchased materials enter the plant through receiving and either go to inspection or, if coded as ‘dock-to-stock’, go directly to the using work cells. At this time, only 30% of the purchased items must be inspected upon receipt. Purchased items are designated as dock-to-stock if the QA department determines that they need not be inspected. The goal is to have all incoming parts as dock-to-stock and thus totally eliminate incoming inspection.
Parts that fail to pass inspection are sent to the material review board (MRB) area. This is a temporary holding area for ‘bad’ material. Here, the decision is made as to either rework the defective parts in-house, use them as they are, or send them back to the vendor.
All other parts are sent directly to the printed circuit board (PCB) production area, the final assembly (FA) work cells, A, B, C and D, or to location Kit. The production of PCBs is the first in-house production step. Components enter this production area, are built into PCBs, and then sent directly to the FA work cells. As such, the PCB area may be viewed as an in-house vendor to the FA work cells.
Work cells A, B, C and D are the final assembly areas. Work cell A produces the Alpha terminals. Cell B produces Headlight terminals and Micro-VA workstations. Cell C is the final assembly work cell for Micro-VA central processing units. Cell D is the Sun work cell. Work cell D is differentiated from the other FA cells in that no in-house manufactured PCBs feed into it. In each FA cell, the mechanical assembly, cable assembly, and the top-level assembly operations for a particular end-item (or similar group of end-¬items) are consolidated. In addition to these assembly operations, the top-level testing, burn-¬in, clean-up and packaging activities are also performed in the FA cells.
Also shown in Figure 5 is location Kit. Kit is a staging area for purchased, shippable products, such as printers and plotters. Kit products will eventually be packaged with the completed final assembly items in finished goods inventory (Fin).
Figure 6 illustrates the production steps that the Alpha terminal would go through. PCB production follows nearly the same sequence of operations as under the old MRP-¬based system. The major process differences include the elimination of the PCB preparation operation, and the elimination of all but one inspection station.
The FA work cell is the consolidation of all the non PCB production related activities necessary to assemble an end-item. The cable and mechanical subassemblies are now built up in the FA work cell, rather than in their own functional areas. Purchased parts making up these subassemblies are delivered directly to the FA cell from receiving (or inspection if not coded as dock-to-stock). Typically, a minimum level (one or two) of these subassemblies is built up one day ahead of time for the next day’s production of end-items. The cable and mechanical assembly processes follow the same sequence of operations as under the old MRP system except for the elimination of all the inspection tasks.
Production scheduling and control
With JIT, production scheduling is no longer done at the subassembly level. Scheduling is done only at the final assembly level. In conjunction with this fundamental change, a completely new production control system was devised.
The first step in developing this new system was to eliminate work orders generated by the MRP system. Since production was only to be scheduled at the end-item level, there would no longer be a need for the lower-level work orders. Instead, a card system, similar to the Toyota Kanban system, was used to control the flow of materials. Under this system a ‘move’ card was used to move material (or assemblies) between successive work centres. ‘Production’ cards were used to build or test materials within a work centre. These cards identified the production lot sizes, move quantities, routeings and so on. With this card system, the practice of picking materials from the stockroom was eliminated. Material inventories were now placed in bins located directly in the work cells. As a result, the central stockroom was eliminated.
The card system, however, was short-lived. It was soon realized that the use of these cards was unnecessary. Having come from an environment characterized by work orders and excessive tracking and control mechanisms, the natural tendency was to use the card system to maintain control over the production floor. Experience showed that the workers could build to the daily production quota and move items to the proper place without the cards telling them what to do. As a result, control of the shopfloor was given to the equipment operators and a new system unique to Applicon’s work environment was developed.
In this system a production card initiates work at the first operation in the production process, the building of PCBs. The PCBs are scheduled on a rate per day basis and are ‘pushed’ through the production floor. The operators at the first process of PCB production, HPDI insertion, read the information on the production cards and build to that schedule. All subsequent PCB processes work to the same daily rate. The PCB production cards (Figure 7) are issued by the production scheduler four days before an order is due in finished inventories. The four days is the total production lead time for a completed unit.
At the other end of the production process final assemblies are ‘pulled’ by customer orders. Using actual customer orders, the production scheduler converts the weekly production plan into specific daily requirements. This information is posted on a status board visible to all the FA work cells. The production scheduler also places end-item build cards in a sequential pile in each corresponding work cell. These build cards (Figure 8) are used to schedule work in the [mal assembly work cells. They identify the customer for which a top-level unit will be assembled, and any specified options that must be built into that particular unit. A separate, individual build card is used for each top-level item. To assemble a terminal, the operators will remove the card from the top of the pile and build the terminal with the specified options. Only one terminal is assembled at a time. Thus, for example, if the status board indicates that two terminals are required for Monday, then the first two build cards on the pile will be used to assemble the customer-specific units for that day.
Under JIT, only two inventory transactions are made. The first transaction is to credit item inventory balances in one general floor-stock location (FLS) upon receipt of materials at receiving. Floor-stock includes all the areas shown in Figure 5 except for MRB and Fin. The second transaction occurs when finished products from work cells A-D, and peripherals from Kit, enter Fin. This transaction is the system “backflush”.
Backflushing is widely used with JIT. Under the old system, inventory tracking was based on detailed transactions that were generated by the MRP system whenever materials (kits) were put into or released out of the central stockroom, or whenever the kits moved from one task to the next in the production process. Eliminating work orders and very short lead times made this system of control obsolete.
With the backflush transaction, the finished products (and any purchased peripherals) are credited to Fin upon entering finished inventories. At the same time, all the lower-¬level components that make up these products are subtracted from FLS. These lower-level components are identified by the MRP system bill of materials. The MRP system has been modified to treat all non-purchased components and subassemblies in the BOM as ‘phantom’ parts. Thus, when backflushing, the system will work through the BOM and stop only when branches that end with purchased items are reached. The appropriate usage multiples of these purchased components are then subtracted from FLS.
Part number: 32820-001
Description: Fr.buffer bd.
Supplied by: Work centre A
Outbound: Wavesolder rack
Figure 7 PCB Production card.
MRP and capacity planning
As a result of these modifications and refinements, MRP is now limited to two major functions. The first is to generate orders for purchasing on a monthly basis.
Using the MRP system, top-level model ‘family’ quantities are translated into MPS purchased parts requirements. This is done through a family bill of materials. An example is shown as Figure 9.
The family bills of top-level items are broken down into 50 and 60 Hz models, with each such model broken down into the various options in which it is sold, based on percentage usage figures. These figures, which are based on historical data, are the predicted fractions of the total family sold as one of the two frequency models, or as some particular option. Thus, for example, we can see from Figure 9 that if 100 units of this model family are forecast to be sold next month, then 70 units will be designated as 60 Hz, 30 units as 50 Hz, and so on.
By using family bills, the master production scheduler only schedules the top-level (family) items. The system uses the percentage usage figures to automatically convert these top-level family quantities into the various specific saleable item (option) requirements.
These requirements are used to create Applicon’s MPS. The MPS is stated in saleable item terms and has standard available-to-promise capability. The MPS quantities then become the new MRP gross requirements when the system is regenerated. The necessary quantities and timings of specific purchased items can then be determined by exploding through the manufacturing bills of material, netting against the gross requirements, offsetting for production lead times, and adding a safety lead time factor.
Figure 10 illustrates how the scheduling of purchased items is carried out in the ideal case. In this example the requirement for the purchased parts is at the beginning of August. Added to this is a two week safety lead time to protect against any uncertainty in the vendor delivery date. Under these conditions, a purchase order must be released such that 1 August items are received on 15 July.
This purchase signal is transmitted to the planner-buyers, who are responsible for material inventories. The planner-buyers make the final decision as to following the recommendation of the MRP system or following some other course of action.
The second function of MRP is to generate capacity plans from the MPS quantities.
These capacity projections are used as a check on labour levels, to determine it there are enough resources at each work centre, and to identify potential bottlenecks.
To generate the capacity plan, the MRP system converts the MPS quantities into corresponding quantities of the various lower-level items. Capacity projections are then calculated by combining these quantities with information from the routeings. The information used from the routeings includes the sequential operations needed to build a particular product, the work centres at which those operations are performed, and the standard production times for each work centre. The capacity projections can be output by individual work centre, by specific part number, or for the total production floor. The capacity reports are run on a weekly basis.
Figure 11 is a sample capacity status report. Here, capacity requirement projections for one month are broken out by individual work cells. The work cell capacity hours include both labour and machine hours. Seventeen work cells are used for capacity planning purposes. The first column in the report identifies the capacity planning work cell. The second column figures are the load hours. These are the projected capacity requirements that are calculated by the MRP system. The load hours can then be compared to the standard capacity hours available, to efficiency adjusted hours (adjusted standard capacity hours), or to maximum capacity hours (standard capacity hours adjusted for 10 per cent overtime).
The MRP system itself is in monthly time buckets. The system is run at the end of each week and the records updated as to that week’s activities. This weekly regeneration ensures that purchased item scheduling and capacity planning are provided with accurate, up-to-date information.
06/12/97 8:12:24 Work centre
Co App Dv
Adjustment percentage 70 Date range:
Work Load Capacity Capacity Capacity
centre (std. hours) (standard) (adj. std) (maximum)
ALF-A 70.03614 480.00000 336.00000 528.00000
ALF-T 5.28740 80.00000 36.00000 88.00000
HLT-A 458.36500 800.00000 560.00000 880.00000
HLT-T 85.08024 160.00000 112.00000 176.00000
MIS-A 270.60698 800.00000 560.00000 880.00000
MIS-T 14.65010 80.00000 56.00000 88.00000
MUX-A 399.27339 1120.00000 784.00000 1232.00000
MUX-T 79.47110 80.00000 56.00000 88.00000
OLD-P 81.10000 0.00000 0.00000 0.00000
PCB-A 52.24560 160.00000 112.00000 176.00000
PCB-H 44.64200 160.00000 112.00000 176.00000
PCB-I 124.91800 320.00000 224.00000 352.00000
PCB-M 441.85760 480.00000 336.00000 520.00000
PCB-P 408.79479 960.00000 672.00000 1056.00000
PCB-T 918.10800 1680.00000 1176.00000 1848.00000
PCB-V 123.24500 160.00000 112.00000 176.00000
PCB-W 56.76000 160.00000 112.00000 176.00000
Totals 3634.44134 7680.00000 5376.00000 8440.00000
20 total workdays included
Figure 11 Capacity status report.
Master production scheduling and order processing
With JIT, the master production schedule is still updated on a monthly basis. However, much less reliance is placed on the forecasts generated by marketing. Today’s approach is to look at the past five or six months’ shipment trends and at the actual orders booked. Don Hathaway, the manager of materials and logistics commented on today’s production planning method: ‘There is little need for any formal forecasting techniques … the approach is to reconcile past trends with marketing expectations, roll up the results, and get an overall plan. The accuracy in estimating overall monthly requirements within 30 days is about 95 per cent, and between 60 and 70 per cent when looking 60 days out.’
The MPS is in monthly time buckets and uses a five month planning horizon. It is stated in top-level item terms and coded by major model number. These top-level models include five processors, sixteen workstations and sixteen major peripherals. The information from the MPS is used only for planning purposes (i.e. to plan and order purchased materials as necessary). With JIT, Applicon plans to forecasts but builds only to actual customer orders. By building to orders rather than forecasts, finished goods inventory levels are kept to a minimum.
Customer orders enter the system from ‘order processing’. Whenever order processing books a new order from sales, or changes the status of an existing order, a ‘work order’ is generated and printed the following morning in the ‘new and modified sales order’ report. The information from this report includes the quantity booked for the new or modified orders, and the due dates of those orders. The only orders that appear on this report are those that have been changed or booked on the previous day.
The information from the daily ‘new and modified sales order’ report is automatically updated into the order backlog records. From this, a weekly ‘manufacturing order schedule/backlog’ report is generated. This weekly report contains all the necessary information on the orders in backlog and links order processing to the shopfloor.
The ‘manufacturing order schedule/backlog’ report is used by the production scheduler to control PCB production. The order quantities from the report are converted into daily production rates. These daily rates are then posted on the PCB production cards. Work on the final assemblies is also scheduled from the information given by the ‘manufacturing order schedule/backlog’ report. In this case, the production scheduler transmits the information about a customer’s order to the final assembly work cell on the end-item build cards.
In September 1997, Bill King, the director of materials, Deborah Cantara, the production control supervisor, and Don Hathaway were reviewing Applicon’s progress towards full JIT implementation.
One issue that needed to be addressed was production control’s suggestion of modifying the MRP system to weekly time bucket. It seemed to Don and Deborah that this was a good idea, but they wondered if the payoffs would be worth the costs. The group also discussed the possibility of completely eliminating finished goods inventory. However, they needed to know what this action would entail and what the ramifications would be.
Another issue that concerned them was how to respond to corporate’s evaluation of the Billerica plant’s manufacturing performance. A new controller at Schlumberger corporate headquarters felt that, amongst other things, the plant had too much excess capacity and that labour and equipment were underutilized. The controller was advocating a standard cost accounting system at Billerica, and Bill King was asked to respond to this suggestion.
The group also wondered what to include in a report to the new vice-president of manufacturing concerning the benefits of JIT. In addition, the vice-president had requested an action plan of recommendations as to what should be done next. Although Bill, Don and Deborah were happy with the results that had been achieved, they wondered what new improvements were possible …. What should Applicon do next?