Our company is ZELTA Manufacturing. At the beginning, our group set up the initial objectives: £500,000 profit and 97% customer satisfaction level. Then some opening strategies were designed to achieve the goals. Above all, our group forecasted the demand for products in level 1. Comparing the average weekly demand of the first 12 weeks of the last year with that of this year, there was a trend with an approximately 15% annual growth in demand for the products. Therefore, our group predicted that the demand of this level would increase 15% as well. According to the average weekly demand of last year (5415), this year aggregate demand was expected to be 6215, (5415 * (1 + 0.15%)). Since the periodicity of peak demand shown by historical data, it was better to schedule production monthly than weekly. So, the forecast of total demand in level 1 was 24860. With taking account of the capacities and costs of the shifts, figure 1 indicated that it was better to use full capacity of the each shift and the day shift was the most profitable among all with lowest unit cost.
Total Cost (£)
Unit Cost (£)
Day shift (3000 - break even)
23000 + 3000 * 10
17.7
Day shift (5000)
23000 + 5000 * 10
14.6
Night shift
22000 + 3500 * 16
22.3
Saturday shift
10000 + 600 * 16
32.7
Evening shift
2600 + 250 * 14
24.4
Day + Night shift
23000 + 5000 * 10 + 22000 + 3500 * 16
17.8
Day + Night + Saturday shift
23000 + 5000 * 10 + 22000 + 3500 * 16 +10000 + 600 * 16
18.7
Figure 1
Sometimes, the actual demand was much smaller than the forecast so that there was high stock of products. Although high production possibly led to inventory holding cost, it reduced the risk of failure to satisfy customer demand which would result in a penalty of 12% of the sales value of under-delivered goods. So our group chose to hold inventory to satisfy customer demand in case. As a result, our group decided to product 5000 (day shift) or 8500 (day shift + night shift) every week and the monthly total amount should be closed to the forecast. E.g. week 13 was set to produce 5000; week 14 was scheduled to be 5000; week 15 was 8500 and week 16 was 5000. Due to cheap unit cost and two weeks lead time, Accessories was classified into Class C with low annual spend by taking Pareto analysis. Hence, it was better to order Accessories in large batches with high levels of safety stock. By contrast, materials such as Main body and Aerial with high unit cost and short lead time was classified into Class A with high annual spend that was ordered in time against holding safety stock. At the end of level 1, our group achieved 100% customer service level and cheapest unit cost of production, while, the high inventory level resulted in low cash flow with the danger of bankrupt, which happed in level 2.
1.2 week 17 - 20
In the following 4 weeks, our group did a new forecasting which worded out 27.8% annual growth in demand by comparing first 16 weeks of last year with this year and this year aggregate demand was expected to be 6920. Since some problems occurred such as scrap, rework and machine broken down, the production was reduced by small amount. Hence, in this month the amount of scheduled production should be nearly 10% larger than demand forecasting. Then our group decides to produce 8500 in week 17 to offset production lost and also to satisfy customer demand in case if the peak demand occurred. After that, the scheduled production in week 18, 19 and 20 was 5000, 8500 and 5000 respectively. Unfortunately, the actual demand of week 17 was 1100, which caused 13000 products in stock and low cash on hand. Our group considered that the stock with 13000 finished goods was large enough to satisfy the peak demand, so the firm produced nothing and paid for holding inventory and fixed costs. That was a big mistake that we did not take account of "break even" volume which could prevent the firm from loss. For instance, if the company uses day shift in production, the fixed cost per week equals £37000 plus £23000. The variable cost is going to be labour cost per unit (£10) plus each Main Body (£8) plus each Aerial (£6) plus Accessories (£1), so each product can make £20 profit (sale value - variable cost). Thus the company needs to produce 3000 products to offset the fixed cost (fixed cost / profit). In addition, the firm that only holds a lot of materials is not able to satisfy customers, so that the firm chose to hold stock with maximum finished goods and minimum raw materials. In week 18, there was still low actual demand. According to the periodicity of demand and forecast, there would be very high customer demand in the following two weeks. Hence our group took a risk that spent all money in production. As predicted, the peak demand occurred in week 19. However, there was no order for materials this week so that the rest of raw materials were only allowed to produce 5000 in week 20. As a result, the stock of finished goods was not enough to satisfy customer demand and then our company got a penalty for failure. To sum up, it was not good to produce the large amount at the beginning of the period and also the end. Taken this month as an example, our firm built large production in first week. It was too early when the big order occurred in third week. Therefore, in the first and second week the high inventory tied up the cash, which made the firm in danger. Also if we built large production in final week, it was too late when big order occurred in week 1 or 2 or 3.
1.3 week 21 - 24
In level 3, a trend with a 29% annual growth in demand was found out by comparing first 20 weeks of last year with this year. Hence we forecasted that this year aggregate demand for standard model was 6985, and monthly demand for standard model was about 28000. Since the demand for XL model was about half of that for standard model, this year aggregate demand for XL model was about 3500 and monthly demand was 14000. With the experience in level 2, our group decided to build big production in middle weeks. Week 21 was scheduled to produce 6000 of standard model and 3000 of XL model by using both day shift and night shift to hold some stock, because the firm paid for penalty for failure to satisfy customer demand last week. Week 22 was set to produce 4000 of standard model and 2000 of XL model using day shift; week 23 used both day shift and night shift to build 10000 products in total and week 24 was scheduled to be 6000. Compared the totally actual demand of week 21 and week 22 with the predicted monthly demand, the total demand for two models during final two weeks would go nearly 30000. As a result, we changed the scheduling of week 24 to be 12700 with the full capacity utilization of all shifts (day shift + night shift + Sat. shift + 5 * Eve. shift). In summary, with the accurate forecast of demand, our group made a success in production scheduling. Firstly we did monthly scheduling on basis of demand forecasting and strategies like minimal raw materials and maximum finished goals. Then according to actual demand and cash on hand, the weekly scheduling is designed to adjust to new situation. At last, our company achieved 97% customer satisfaction and made £386k profits, which was closed to the initial goal.
2 Master Production schedule
2.1 week 25 - 28
In this section, it looks for producing a Master Production Schedule for weeks 25 - 35. Owing to the periodicity of peak demand, it is better to design the schedule month by month, so firstly it is going to work for week 25 - 28. Above all, it needs to forecast the demand for two models. The actual demand for standard model during the first 24 weeks of last year is 113099 and that of this year is 132100. With comparing the demand of two years, there is a 16% growth in demand for standard model. Therefore, it forecasts that this year aggregate demand is 6281, (5415 * (1 + 16%)). Since the XL model is in the introduction phase of its life cycle, there is unsharp change in the demand for it and this year aggregate demand for XL model is about half of that for the Standard model: 3100. Secondly, it is going to design the quantity of production and material used in each week. The number of Standard model and XL model left from last month is 1353 and 788 respectively. With the small amount of stock of finished goods, it is going to produce as many as possible in week 25 in case of peak demand. In contrast with scheduled production, the actual production is found to be 89.7 per cent of it on average. In addition, the numbers of Main Body and Aerial are both 4875 and the number of Accessories is 42998. For this reason, the maximum quantity of material used in production is only 4875. Through the calculation of the material usage in the last 12 weeks, it is found that the quantity of Main Body and Aerial used in production is 1.08 times the actual production while the Accessories is 1.44 times the actual production. Hence, it sets the scheduled production of week 25 to be 4500 involving 3000 of Standard and 1500 of XL as showed in figure 2. The money used in production in week 25 is going to be £137000. Fortunately, the cash at the beginning of week 25 is £329303 that is enough to afford production cost.
As mentioned above, it is more profitable to use day shift and the full capacity, so the week 26 is schedule to product 6000 (day shift) or 10000 (both day shift and night shift). If it produces 10000, the cost will be £395000. Therefore the actual demand in week 25 should be at least roughly 3000 for Standard and 1500 for XL to afford the production in week 26. According to historical data, it is unlikely to sell 3000 of Standard model and 1500 of XL model in the first week of the month, so it is going to produce 6000 using day shift only in week 26 and then the cost will be £242000. It only needs to sell about 1000 of Standard and 500 of XL in week 25 to pay for the production, which is very easy to achieve. In the other side, the actual demand is unknown so that it is unable to measure cash on hand. Generally the peak demand appears in the third week. Also compared the scheduled production with the forecasting, it plans to build 10000 products to avoid penalty by using both day and night shift. Finally, in order to match the forecasting demand and keep high customer satisfaction level, it is designed to apply full capacity utilization in week 28.
Week
25
26
27
28
Standard
XL
Standard
XL
Standard
XL
Standard
XL
Scheduled production
3000
1500
4000
2000
6700
3300
8300
4400
Material ordered
Main Body
6000
10000
12000
6000
Aerial
6000
10000
12000
6000
Accessories
0
0
0
10000
Figure 2: Schedule of week 25 - 28
Furthermore, the quantity of material ordered in this week is based on the scheduled production of next week. Due to high unit cost and short lead time, Main Body and Aerial are ordered week by week to avoid holding stock. Therefore, in week 25 it only orders 6000 of Main Body and Aerial and zero for Accessories, because it is designed to have 6000 products in week 26 and already has nearly 50k Accessories in stock shown as figure 3. At the end of week 28, there are enough Accessories in stock even it is going to use full capacity of shifts in week 19. Since the cheap unit cost and two weeks lead time of Accessories, it orders 10000 of Accessories for week 30 in case.
Week
25
26
27
28
Standard
XL
Standard
XL
Standard
XL
Standard
XL
Actual production
2691
1345
3588
1794
6010
2960
7445
3946
Material used
Main Body
2916
1458
3888
1944
6512
3207
8067
4276
Aerial
2916
1458
3888
1944
6512
3207
8067
4276
Accessories
2916
2916
3888
3888
6512
6512
8067
8067
Balanced
Main Body
501
669
1197
854
Aerial
501
669
1197
854
Accessories
47166
46390
33366
17232
Figure 3
2.2 week 29 - 32
Then, the following section is talking about the scheduling of week 29 - 32. It is found that there is a 10% reduction in demand for Standard model through comparing the actual demands of week 17 - 20 and week 21 - 24. Therefore, it assumes that the total demand for Standard model in the last month is roughly 20000 and there are only 1000 of Standard and 700 of XL in stock at the end of week 28. The smaller quantity of demand for Standard indicates that Standard model is approaching the end of its life cycle. Hence, there is an 8.7% annual growth in demand for Standard and the forecasting demand is 5886. The XL model succeeds in its introduction phase and moved to growth phase. Then the demand for XL model increases at a rapid rate. It assumes that there is a 10% increment in demand each period. Hence, it will cost nearly half to one year for XL product to be mature. Also it is important that the capacity is able to keep pace with the increased demand to prevent the waste of sales. In this period, the demand for XL is predicted to be 11000 and the more capacity is allocated to the production of XL model than previous two periods. Since it is too early to build large production in first week when the big order occurs in the later week, it is designed to produce 6000 in week 29, 10000 both in week 30 and 31. The large production built in middle of the month is the best way to avoid peak demand. Then the week 32 is scheduled to produce 6000 products by only using day shift which is the most profitable. The total production in this month is so closed to the forecast.
Week
29
30
31
32
Standard
XL
Standard
XL
Standard
XL
Standard
XL
Schedule production
4000
2000
6000
4000
6000
4000
4000
2000
Material ordered
Main Body
9000
10000
6000
6000
Aerial
9000
10000
6000
6000
Accessories
20000
10000
10000
20000
Material used
Main Body
3888
1994
5832
3888
5832
3888
3888
1994
Aerial
3888
1994
5832
3888
5832
3888
3888
1994
Accessories
3888
3888
5832
7776
5832
7776
3888
3888
Balanced
Main Body
1022
302
582
750
Aerial
1022
302
582
750
Accessories
9456
5848
2240
4464
Figure 4: Schedule of week 29 - 32
2.3 week 33 - 35
It assumes that there is a 10% decrease in demand for Standard products each period because it approaches the declining phase of its product life cycle. Then the forecasting demand for Standard model and XL model is 18000 and 12100 respectively. In the declining phase, the inventory must be managed to avoid obsolescence and also penalty to dissatisfied customers. In order to hold appropriate stock level of finished goods, it sets week 33 to produce 6000 products with the most profitable day shift. In case of the peak demand occurring in later weeks, it is designed to build large production in week 34. Then it chooses to produce 10000 in week 35 to satisfy customer demand. Also it orders enough materials for the next week.
Week
33
34
35
Standard
XL
Standard
XL
Standard
XL
Scheduled production
4000
2000
6000
4000
6000
4000
Material ordered
Main Body
10000
9000
10000
Aerial
10000
9000
10000
Accessories
10000
10000
10000
Material Used
Main Body
3888
1944
5832
3888
5832
3888
Aerial
3888
1944
5832
3888
5832
3888
Accessories
3888
3888
5832
7776
5832
7776
Balanced
Main Body
918
1198
478
Aerial
918
1198
478
Accessories
6688
13080
9472
Figure 5: Schedule of week 33 - 35
3 The end of the Standard product life cycle
Since the Standard product is not dependent upon fashion, production cannot be stopped with little or no problems (WMG). Therefore, the Standard model will be phased out. At the same time, a new product should be introduced to replace the declining products. The improved product such as XL model is sold at a high price to sustain the sales as the Standard product approaches the end of its life cycle. For this reason, cannibalization is a good way to gain and defend the market share (Komninos, 2002). The demand for XL model will increase rapidly instead of Standard model. Above all, even the demand has shrunk to a fraction of its peak, the company still attempts to milk all remaining profits from product. Firstly, it maintains a high price policy for the declining product that increases the profit margin and gradually discourages the loyal customers from buying it. Also the company focuses on cost reduction that withdraws capital expenditures in promotion of Standard product while there is a big increment in advertising XL model to encourage customers purchasing it. In addition, although the demand for Standard product is decreasing, production equipment has to be kept running until the most profitable part of the life cycle has ended. As a result, this is time to start withdrawing variations of the Standard product that are not profitable in the market and abandon most channels of distribution except basic channels used in the development phase. Moreover, it is better to order the material in time than hold spare in stock, because inventory must be managed to avoided obsolescence. Owing to the same materials used in both Standard model and XL model, it is not a big deal to hold high level inventory of materials. However, the Standard product is held as few as possible in stock to minimise waste, which requires accurate forecast for demand. In the other side, the Semi-fixed costs and the Normal capacity apply to both the Standard and XL model. It means the declining product and the improved product can share the same production facilities. Hence, there is no need to sell some equipment or fire part of workers after the Standard product discontinued. The company moves the centre of production from Standard model to XL model so that the majority of facilities and work force are used to produce XL product. Finally, since the demand for Standard product is declining, the amount of production is also decreased accordingly. Hence, it cannot achieve the scale of economy any longer so that the production is less profitable. If the cost of production is closed to the revenue of Standard product, there is nearly no profit to product it. As a result, the assembly line of Standard product is force to terminate.
4 Relationship between customer service, inventory and forecasting
Customer service is defined as "the right person receives that right product at the right place at the right time in the right condition and at the right cost" (Murphy and Wood, 2008). Demand forecasting refers to efforts to estimate the product demand in a future time period and forecasting accuracy is the difference between actual and predicted demand. Inventory refers to stocks of goods that are maintained for a variety of purposes such as to support manufacturing or customer satisfaction. One of the key classifications of inventory is safety stock that holds against uncertainty in demand. The level of safety stock is affected by three factors: the forecasting accuracy, the desired customer service level and re-supply lead-time.
Above all, inventory holding levels are affected by the desired customer services. In the worst case, the shortage will impact on the service given to the customer incurring potential penalty costs. If there will be shortages, the firm has to add safety stock to fill orders and improve the service level. In practice, there are two dimensions of customer service related to inventory level: time and reliability. First, the long replenishment lead times result in high inventory requirements to avoid failure. The reason is that if demand during the replenishment lead time is higher than available inventory, then unsatisfied demand is lost which in turn can negatively affect company's reputation and future demand. Therefore, companies are looking for ways to reduce lead times and inventory level. In addition, the inconsistent order cycles, loss and damage in delivery can reduce the reliability of service and then cause some negative effects such as out of stock situations. Generally it would be better to use safety stock when there is uncertainty in the demand. Nevertheless, if the demand is less than the available inventory, the excess stock will cause additional inventory holding cost. Therefore, the reliable demand enables a firm to maintain a lower level of inventory, which produces lower inventory holding costs (Murphy and Wood, 2008).
Furthermore, the forecasting accuracy plays an important role in managing inventory and customer service. As mentioned above, the quantity of the replenishment of inventory is likely to be forecast. It is well known that the more forecasting accuracy, the better supply and demand match up and the lower the inventory level holds. For instance, Sunbeam changed monthly demand forecasting to include demand estimates of their top 200 customers. This information increased forecast accuracy, which in turn led to a 45 per cent reduction in inventory investment (Webster, 2008). In order to improve the forecasting accuracy, there are some alternatives to change the way the firm operates. To start with, as the forecast horizon increases, forecast accuracy decreases. This means that the longer the lead time, the longer it needs to forecast, so the more unreliable the forecast is. Therefore, it aims to cut down lead time in production and delivery. Taken Dell as an example, it only takes less than two hours after receipt of the order to assemble a specified computer. Hence, instead of developing inaccurate demand forecasts and holding inventory of thousands of possible PCs, the company produces in efficient responds to customer orders and so the uncertainty of demand is reduced. In the other side, it implies that the short-term forecasting is more accurate than long-term forecasting, because the errors will be high if the recent week demand is used to predict the demand a year or more from now. So the demand should be monitored regularly and the period of information that is used for forecasting and decision making should be kept short to improve forecasting accuracy. However, according to different purposes, the firm can select short, medium or long term forecasting.
Secondly, the forecasting accuracy is also affected by the demand volatility, so the firms should change their behaviours and policies to reduce the demand variance. In most cases, variable pricing and promotions can cause the variation of demand. As a result, the firm has to increase inventory level against failure. What is more, if the company has a wide range of products, demand for each product will become more volatile. Sometimes, many categories of products only contribute small percentage of the volume and sales. With taking Pareto analysis, these products can be identified as "C" items and some products accounting for majority of sales can be classified as "A" items (WMG). One research (Webster, 2008) suggests that if the volume of low-value-added items is declined to eliminate demand uncertainty, forecast accuracy will be improved effectively and the impact on sales will be minimal. Hence, the company should narrow the product line and focus on A items to supply high levels of services.
There is an application of the good relationship between customer service level, forecasting and inventory management: Vendor Management Inventory (VMI) maintains lean stocks by subcontracting of material control. In the process, the supplier forecasts and manages raw material based on the demand information provided by the customer. The information exchange increases the understanding of market trends and customer tastes, ultimately reducing inventory and improving customer service. By contrast, if the company has the poor performance in forecasting, the consequences of inaccuracy forecasting are either shortage, lost sales and dissatisfied customers or excess stock. For example, Apple had poor demand forecasting in 1995, which led to $ 1 billion unfilled orders in the second quarter of 1995 (Caroll, Carlton, and Rigdon, 1996).
5 LEAN/JIT in service
JIT approach is not only applied to manufacturing industry but also service industry. Since services are simultaneously consumed as they are produced and cannot be inventoried, it is difficult to match the capacity with demand. As a result, there is always imbalance in supply and demand that can result in either idle employees and resources if demand is smaller than supply, or lost sales if demand is larger than supply. The objective of JIT approach is to eliminate waste such as idle work force or sale loss so that it can be described as a low cost/high return policy. Hence, the company with JIT should decide appropriate capacity of service to match variable demand and avoid waste. There are many strategies used to manage supply and demand in different types of service. For instance, if customers are not physically present, the companies try to reduce the use of service employees through substitution of technology for people such as automated phone answering systems with simulated voice. Otherwise, if customers directly interact in the service transaction, creative work schedule such as nonuniform times is flexible to cope with variable demand. The variable work hours not only can satisfy the peak demand but also maximize capacity utilization. In addition, a company can shift cross-training personnel temporarily to increase the capacity of the service. However, it is too expensive to cross-train personnel for different jobs.
Taken the airline as an example, it can use reservation to match their capacity to demand. Since customers are fickle and so not always show up, airlines always adopt overbooking system to compensate for no-shows and minimise empty seats on the flights. Overbooking refers to selling of a volatile service in excess of actual capacity. Thus, the airline has to balance the risk of a no-show with the compensation they have to pay to bumped customers. Nevertheless, it is difficult for the company to determine how much to overbook. There are three approaches to calculate the overbooking level. Firstly, the number of overbook is set to be the average of no-shows. Secondly, the level takes account of the comparison between the costs of no-show and overbooking. The last is marginal cost approach that keep accepting bookings until the expected revenue is less than or equal to the expected loss from the last booking (Metters, King-Metters, Pullman and Walton, 2006). In the other side, the airline often holds capacity for more profitable customers. Then it charges higher price to price-insensitive customers and lower prices to capital-intensive vacationers. According to the sensitivity of price, the demand by vacationers can be altered by price changing. Hence, higher price can flatten demand peak to conform to capacity of service, while cheaper price attracts more customers to avoid empty seat. Even though the overbooking sometimes causes an oversale in economy class, the company can upgrade some passengers to higher classes. In practice, if the demand is much larger than the capacity, the company can add additional air flights to transport passengers. Moreover, information systems play an important role in JIT to match supply and demand, because the more accurate information is used in calculation, the better combination of overbooking level, capacity segment and price is worked out.
