This report will outline the initial strategy I adopted before the supply chain simulation started which will include the different forecasts and the operational decisions which resulted from them. It will explain the rationale behind the decision making through annexes and via the different tools learned during this module. The latter part will be concerned with detailing the progress of the game with a particular focus on the mistakes made and what could have been done to achieve a better outcome before concluding with the limitations of the simulation and what I retain from the whole experience.
1. Strategy
I have decided to supply all markets while ensuring I adjust the relevant variables so that the profit margin achieved in any geographical location remains the highest possible. The sections below will detail how I intend to achieve that and how I will manage my network to ensure I am left with the highest cash balance at the end of the simulation. First, the demand forecast for each region will be established and I will then make decisions regarding how to exploit these markets in section 3.
2. Demand forecasting
2.1. Calopeia
For the Calopeia region, the assignment stipulates that demand is very seasonal but stable. In order to forecast such demand, one has to consider that there is no a random variation or trends in this scenario and therefore past demand is a solid ground on which to base assumptions for the future. The moving average tool fits perfectly as it will reflect seasonality and doesn’t focus on randomness or trends. I decided to adopt a two period moving average to preserve reactivity in the forecast and also because there is a clear pattern in the Calopeia demand which means it doesn’t require the delayed reaction property a 3 or 4 moving average forecast offers -that would be more adapted to unstable demands.
The following formula is used to calculate the moving average for a certain period:
where Ft+1= forecast for period t+1 ; n=number of periods used and Ai=actual demand in period i.
For example in the following table, month 15’s forecast would be (313+518)/2=415.5 ; month 16’s forecast would be (518+878)/2=698 and so on.
Month
|
Orders
|
Two Period Moving Average Forecast
|
13
|
313
| |
14
|
518
| |
15
|
878
|
415.5
|
16
|
1184
|
698
|
To consult the full table and the corresponding graphs, please refer to Annex 1 which shows total expected demand for this region is around 27 778 units.
2.2. Sorange
For the Sorange region, I am told that although the size of the orders are random, overall demand will grow linearly until day 1430. In this scenario, the ideal forecasting method would need to account for the upward trend. In this regard, the Linear regression method is best suited. If I summarize the order data available slightly after day 730 by month, I obtain the following table:
Month
|
Orders y
|
Time Period x
|
x2
|
xy
|
22
|
27
|
1
|
1
|
27
|
23
|
180
|
2
|
4
|
360
|
24
|
281
|
3
|
9
|
843
|
25
|
441
|
4
|
16
|
1764
|
Totals
|
929
|
10
|
30
|
2994
|
I now need to determine the intercept (a) which is given through the following formula:
where n is the number of periods.
a= (30*929 - 10*2994) / (4*30 - 102) = -2070 / 20 = -103.5
I can now determine the slope (b) with this formula:
b= (4*2994 - 10*929 / (4*30 - 102) = 2686 / 20 = 134.3
Now that I have the values of a and b, the following regression equation can be used to forecast future months’ orders:
Y= a + bX = -103.5 + 134.3X
With this equation, I could construct the full forecast table and the corresponding graph in Annex 2 where one could see that estimated remaining demand for Sorange is 51 629.
2.3. Tyran & Fardo
Tyran and Fardo’s demands behave differently from the first two markets. Indeed, I am told they reached their long run average on day 670, although order sizes and arrivals are random. Because neither market will show any trend or seasonality but are both random in terms or order size and arrival, I decided to adopt the exponential smoothing technique with a constant of 0.3 to reflect the considerable randomness in these time series. The formula for this tool is as follows:
Ft+1 = αDt + (1 - α) Ft
Where Ft+1= new forecast, Ft=current forecast, Dt=value for time period t and α=smoothing constant 0.3
Note that because demand for both regions stabilizes at month 23, I will not consider month 22. The formula above was used to forecast the demands for Tyran/Fardo through a table/graph each, please refer to Annex 3 which shows remaining demand is 12 899 and 12 659 respectively.
2.4. Entworpe
Considering Entworpe, demand is not seasonal nor is it trending and orders of 250 units arrive randomly. The first four months of demand are summarized below:
Month
|
Orders
|
22
|
250
|
23
|
500
|
24
|
250
|
25
|
750
|
After reflection, it was decided that an educated guess was perhaps most appropriate for this particular market. The average demand for these four months is 437.5. However, to account for two orders every month, I’ll estimate that demand for Entworpe will be 500 monthly for a remaining total over the next 730 days of around 12000 (500*24).
3. Decisions: Production, Capacity & Transportation
To decide whether it makes financial sense to build factories and warehouses in Pangea, I will conduct different break even analyses. First, it is useful to break down the costs associated with my activity (based on a batch of 200):
Net Profit per Unit
| ||||||
Batch of 200
|
Same Region
|
Different Regions
|
Factory in continent Warehouse in Fardo
| |||
Mail
|
Truck
|
Mail
|
Truck
|
Mail
|
Truck
| |
Revenue
|
1450
|
1450
|
1450
|
1450
|
1450
|
1450
|
Drum cost
|
-1000
|
-1000
|
-1000
|
-1000
|
-1000
|
-1000
|
Batch cost
|
-7.5
|
-7.5
|
-7.5
|
-7.5
|
-7.5
|
-7.5
|
Inbound logistics
|
-150
|
-75
|
-200
|
-100
|
-400
|
-225
|
Other expenses
|
-3
|
-3
|
-3
|
-3
|
-3
|
-3
|
Holding Cost
|
-2.78
|
-2.78
|
-2.78
|
-2.78
|
-2.78
|
-2.78
|
Outbound logistics
|
-150
|
-150
|
-150
|
-150
|
-150
|
-150
|
Net Profit per Unit
|
136.72
|
211.72
|
86.72
|
186.72
|
-113.28
|
61.72
|
I will use these figures to assess where to build facilities. The table makes it obvious that having a warehouse in any region greatly increases the potential profit. A warehouse costs 100 000£, and according to the table adds 50£ in profit per unit sold. Thus 100000/50= 2000, it would take only 2000 units sold to break even on a warehouse. Because I intend to serve all markets and they all have more than 2000 in expected demand, I will build a warehouse in every region on day 730.
3.1. Calopeia
The Calopeia region disposes of a factory with 70 drums a day capacity and a warehouse. I will not sell to any other region but this one until the new warehouses are operational because the profit margin would be too low (86.72£) - the original stock has been transported via mail. Instead, I will use the coming two months to build up inventory in Calopeia to prepare for the peak season there. Secondly this would allow the Calopeia factory to feed the new warehouses without worrying about running out of stock in Calopeia from month 27 to 29. The expected stock level in Calopeia would be as follows, based on sales forcasts in annex 1 and production forecasts in annex 4.
Calopeia Warehouse
| |||
Month
|
Stock
|
Outgoing
|
Incoming
|
25
|
2528
|
313
|
2000
|
26
|
4247
|
518
|
2000
|
27
|
5694
| ||
The consequent 5694 in stock would be enough to sustain demand in Calopeia until month 30. Although I do intend to resume deliveries to that warehouse from month 29, this would leave me two full months to fill the other 3 warehouses. Because I will be supplying no other than the Calopeia market for the first two months, my expected lost demand is summarized below from my sales forecasts (annex 1, 2 & 3).
Lost Demand
| ||||
Month
|
Sorange
|
Tyran
|
Entw
|
Total
|
25
|
434
|
580
|
500
|
1514
|
26
|
568
|
541
|
500
|
1609
|
27
|
702
|
531
|
500
|
1733
|
Total
|
1704
|
1651
|
1500
|
4855
|
So I expect a maximum of 4855 units in lost demand, which would represent a lost profit of 4855*136.72= 663 776£. I used the figure 136.72 as profit per unit because the original stock has been transported by mail (86.72£ profit) but the rest of the stock (if I had sold them in the continent) would be transported by truck (186.72£ profit) so I simply computed an average profit per unit. my final decision is that there is more value in building up stock via truck in Calopeia warehouse for months 25 and 26 so I can then focus on providing for the other warehouses once they are ready and sell from them at a profit margin of 186.72£. Thus, from month 27, Calopeia factory will serve all my mainland regions. I designed the capacity set up below which I will explain further.
Calopeia Factory
|
Capacity
|
Additional Capacity
|
Additional Units Produced
|
Max monthly production
| |||
Day 730
|
70
|
0
|
51 100
|
2000
| |||
Day 820
|
100
|
+30
|
19 200
|
3000
| |||
Day 910
|
120
|
+20
|
11 000
|
3600
|
Investment for extra capacity (£)
|
Expected profit: Profit per unit* units sold
|
Return on investment
|
10 235 842
| ||
1 500 000
|
3 585 024
|
139%
|
1 000 000
|
2 053 920
|
105%
|
Expected profit has been computed with the figures from the net profit table above and from the forecasts on annex 1 and 4. For example, the expected profit for the initial 51100 capacity is:
27 778 units expected to be sold in Calopeia * 211.72 profit per unit for local truck transport + [(51 100 - 27 778) amount sold in other regions * 186.72 profit per unit for continental truck transport]
27 778*211.72 + [(51 100 - 27 778)*186.72] = 10 235 842£
Those two extra capacity additions would allow the firm to keep up with the Calopeia, Sorange, Entworpe and Tyran demands as is detailed in annexes 4 to 7. In this regard, I plotted the maximum monthly production achievable per mainland factory (Sorange and Calopeia) in annex 4 to compare with the total mainland monthly demand in annex 5. Annex 6 plots production against demand. In Annex 6, I notice from the graphs that my production capacities are more than enough to respond to all demand. I decided not to reduce those capacities for two reasons. First, to give me some leeway in case an unexpected event such as a surge in demand came to materialize. Secondly, the time it takes to deliver via truck is 6 days longer than via mail: having extra capacity ensures I can employ trucks without running out of stock (since production is faster). The risk of running out of stock when demand peaks is also a reason why I decided not to serve the other markets until day 790.
3.2. Sorange
Regarding the Sorange region, I know from Annex 2 that I expect total demand from this country to be around 51 629 total. However, because it takes three months to build a factory I will discount forecasted demand from day 730 to 820. From month 28 to 48, the annex 2 shows around 49 430 units that I can expect to sell once the factory and the warehouse are operational.
Assuming an initial factory capacity of 30, I can now conduct a break even analysis.
Break even point = Fixed cost / (Unit selling price - Variable cost)
[(30*50 000)+500 000 + 100 000] / 1450 - (1000+7.5+75+3+2.78+150)
2 100 000 / 211.72 = 9919
I would need to sell 9919 drums to break even on the initial 2 100 000£ invested, which shows it makes sense to build a factory and a warehouse in Sorange as expected remaining demand there is close to 50 000. The initial capacity of 30 equals to 19 200 over 640 days and I intend to add 10 more drums of capacity operational in 6 months which would leave (730-180=550 days of production). The expected payoffs are:
(30*640*211.72)-30*50 000-500 000- 100 000= 1 965 024£
(10*550*211.72) - 10*50 000= 664 460£
Sorange Factory
|
Capacity
|
Additional Capacity
|
Additional Units Produced
|
Max monthly production
| |||
Day 820
|
30
|
19 200
|
900
| ||||
Day 910
|
40
|
+10
|
5 500
|
1 200
|
Investment for extra capacity (£)
|
Expected profit: Profit per unit* units sold
|
Return on investment
|
10 235 842
| ||
1 500 000
|
3 585 024
|
139%
|
1 000 000
|
2 053 920
|
105%
|
* including Factory 500k and Warehouse 100k.
Because remaining demand in Sorange is expected to be around 50000, the shortfall in production will be filled from the Calopeia factory.
3.3. Fardo
Fardo is tricky as the cost breakdown table shows. Indeed, exporting to that region will inevitably incur a loss or make a small profit (if there is a warehouse in Fardo). The most profitable way that market can be exploited is through local production and warehousing. Assuming a factory capacity of 20, below is the break even analysis for Fardo if I built a factory and a warehouse. Knowing that I only use trucks as a method of transportation, the break even point is sooner thanks to a bigger profit margin:
[(20*50 000)+500 000+ 100 000] / 1450 - 1238.28 = 7557 units.
Fardo Factory - truck transport
|
Demand forecast (annex 3)
|
Investment £
|
Profit per unit
|
Sales
| ||
Factory 20 capacity +warehouse
|
11 272
|
1 600 000
|
211.72
|
2 386 508
| ||
Warehouse in Fardo
|
11 606
|
100 000
|
61.72
|
716 322
| ||
Warehouse in Mainland
|
12 659
|
0
|
-113.28
|
-1 434 012
|
Expected total profit
|
ROI
|
Net Profit
|
2 386 508
|
49%
|
1 080 163
|
716 322
|
616%
|
616 322
|
-1 434 012
|
-1 434 012
|
I scheduled 20 drums a day production because for 640 days of operation it would equal to 12 800 drums, which is enough to cover the expected remaining demand of 11272 for Fardo (See Annex 3). Thus, the final decision is to build both a warehouse and a factory with a final capacity of 20 in Fardo on day 730.
3.4. Tyran and Entworpe
A factory investment in Tyran or Entworpe is not warranted. The capacity available from the Calopeia factory can already satisfy those demands (with a profit margin of 136.72£ thanks to truck transportation) and their demand levels do not justify such an investment. The next section summarizes part 3 of this paper.
4. Summary
The next two tables show that my full production capacity will equal 118 800 units while the total expected demand from all regions will be 116 965 units from day 730.
Production
(additional daily capacity)
|
Calopeia Production
|
Sorange Production
|
Fardo Production
|
Total Production
|
From Day 730
|
51 100
|
0
|
0
| |
Addition from day 820
|
19 200
(+30)
|
19 200
(+30)
|
12 800
(+20)
| |
Addition from day 910
|
11 000
(+20)
|
5 500
(+10)
| ||
Total
|
81 300
|
24 700
|
12 800
|
118 800
|
Estimated remaining demand (see Annexes 1,2 &3)
|
Calopeia
|
Sorange
|
Tyran
|
Entworpe
|
Fardo
|
27 778
|
51 629
|
12 899
|
12 000
|
12 659
| |
Total
|
116 965
| ||||
Thanks to these tables and my previous figures I can roughly estimate my final cash figure on day 1460. First below is a table summarizing my investments in infrastructure.
Investments
|
Factories
|
Warehouses
|
Capacity
|
Total
|
Calopeia
|
2,500,000
|
2,500,000
| ||
Sorange
|
500,000
|
100,000
|
2,000,000
|
2,600,000
|
Tyran
|
100,000
|
100,000
| ||
Entworpe
|
100,000
|
100,000
| ||
Fardo
|
500,000
|
100,000
|
1,000,000
|
1,600,000
|
Total
|
1,000,000
|
400,000
|
5,500,000
|
6,900,000
|
Finally, I will conclude this chapter with an estimate of my final cash figure with the table below:
Calopeia Factory
|
Market
|
Profit Margin
|
Demand
|
Total
|
Calopeia
|
211.72
|
27,778
|
5,881,158.16
| |
Sorange
|
136.72
|
26,929
|
3,681,732.88
| |
Entworpe
|
136.72
|
12,000
|
1,640,640
| |
Tyran
|
136.72
|
12,899
|
1,763,551.28
| |
Sorange Factory
|
Sorange
|
211.72
|
24,700
|
5,229,484
|
Fardo Factory
|
Fardo
|
211.72
|
12,659
|
2,680,163.48
|
Total
|
116,965
|
20,8767,298.8
| ||
Investments
|
-6,900,000
| |||
Holding costs
|
-1,000,000
| |||
Starting Cash Balance
|
6,795,100
| |||
Interest 10%
|
2,000,000
| |||
Final Cash Balance
|
21,771,829.8
| |||
6. Analysis of mistakes and solutions suggestion
The network simulation went rather poorly for this group because of grave divergences from the original plan, some functional mistakes and a general lack of discipline which is perhaps due to the fact that I did not have the opportunity to conduct the first simulation. Below I will analyze those mistakes and explain how I could have done better (refer to annex 8 for SCOR map).
6.1. Inventory issues
On day 730, I changed the transportation method to truck as planned but a typing error meant I ordered 2600 drums on day 740, instead of 1000. This meant that I had to wait 44 days before receiving my order. Disconcerted by the resulting 2.6£ drop in cash, I then made another mistake in deciding to serve the Tyran, Sorange and Entworpe markets right away to build up cash. It lead me to run out of stock on day 762 until 785. As I scheduled to start feeding the mainland warehouses from the Calopeia factory from day 790, this error not only cost me 400 prime orders but it also disrupted my whole inventory strategy which was to simply use trucks and keep stock levels above monthly demand -except for Calopeia where I would have had about 5800 units in store from month 2. Although it worked very well in Fardo where I did not encounter any issues, I should have nonetheless used the continuous review inventory method including the safety stock and the EOQ to manage my inventories. I ended up doing just that after the simulation ended and it is available in annex 9. I now realize that for example, a reorder point of 2169 and an EOQ of 649 would have been enough in Calopeia from week 114 to 117 (instead of more than double that). However, although the continuous review system and the EOQ would have certainly worked very well for Fardo, Entworpe and Tyran, I think the seasonal factor in Calopeia and the trending demand in Sorange would have put those tools into difficulty. Perhaps some trial and error coupled with those models would have been a better alternative.
6.2. Scheduling factories/capacity too late
Initially, I intended to have all scheduled capacity to be operational by day 910 as part 4 specifies but in the simulation I achieved that by day 1026. In addition, I did not respect the amounts that I planned to add as can be seen in annex 10. Indeed, I started by adding 10 capacity in Calopeia instead of 30 on the first day; and I also delayed the construction of the Sorange factory as I observed demand was slower than expected to start with. In hindsight, this was a mistake as my lost orders of 27 149 units attest. That lost demand equals to 3 926 686£ in additional profit (2865*211.72+24284*136.72). Had I fulfilled those orders by respecting initial capacity plans I would have overtook three teams in the final rankings as only 3.8£ more would have been necessary to occupy the 5th place instead of my current 8th spot with 16,432,781.70£.
6.3. Transportation
I also severely deviated from my initial transportation strategy which was to stick to trucks throughout the game. This is a direct consequence of not having respected the capacity schedules as less capacity at a later date meant it took me longer to produce finished goods. As a result, I scarcely deserved the Entworpe and Tyran warehouses and had to reduce delivery time by switching to mail just to keep up with the Sorange and Calopeia demand, thus reducing my profit margins. Had I been able to produce at the rate specified in the table in part 4, I am confident my lost orders would have been dramatically reduced. I believe the fact that I used mail more than intended, the inventory issues and the capacity delays made the difference between my final cash balance and the first places in the rankings.
7. Simulation Limitations
Although this simulation was very helpful at understanding the course content, it does present several limitations.
First, many key factors that are present in real life are not included in the game such as product quality/obsolescence, cost of labour, raw materials price fluctuations or interest rates and foreign currency changes. All these factors could have a major impact on a supply chain. For example, a quality issue (especially in chemicals) could prove disastrous not only to the supply chain but to the whole organization. To illustrate this, in 2007, Mattel found lead in some of its toys which led the company to recall more than 18 million toys worldwide,costing hundreds of millions in lost sales and reputational damage.
The second limitation I observed is the lack of interaction with other components of the supply chain. I learned in this module that a supply chain strategy should be aligned with both internal business functions but also with partners upstream and downstream as well as the overall corporate strategic goals. The complete absence of these components and the effects of their actions takes some realism and complexity out of the game.
Another limiting factor is the lack of an environment. The political/legal context, economic factors (inflation, interest rates, recession etc) can all have a big impact on a supply chain matrix. For example, PwC believes that correctly managing macroeconomic risk in today’s increasingly complex supply chains could make the difference in this fast paced and global economy. The natural environment can also be a source of danger or opportunity. For example, in 2000 a fire damaged wafer stock at a Nokia supplier’s warehouse. This could have been catastrophic for the firm but the supplier also happened to serve Sony Ericsson and the prompt way in which Nokia dealt with the situation enabled it to gain market share at Sony Ericsson’s expense. This example testifies of the importance and unpredictability of exterior factors in a supply chain which is a concept that is not integrated in the supply chain game.
Another omission is the absence of a real Bullwhip effect. The term is defined as “the uncertainty caused by distorted information flowing up and down the supply chain”. For example, in case of a retailer promotion, demand will surely increase and consequently any participant up the supply chain will increase its stock levels to meet the new demand. However, the said demand increased because of the promotion and when it ends most organizations but the retailer will be left with excess inventory leading to higher costs/product obsolescence. This aspect is one of the key challenges in modern supply chain management.
Lastly, the lack of flexibility in the game is another weakness. Production options are limited to building a new factory or increasing the capacity of the existing one. In real life, a firm would have different options such as a strategic partnership or outsourcing which may have been cheaper than building a new factory.
8. Conclusion
In hindsight, this simulation was an excellent way to put the concepts learned during the module into practice. Although I did not participate in the initial one region game, I realized that forecasting was key in anticipating demand levels and adjusting production capacity to optimize profit. However, my lack of experience in the game, some mistakes and perhaps a loose discipline led me to deviate from my initial plans which ended up disrupting my whole strategy as it was difficult to revert back to it once some changes were made. However, I reflected on the mistakes made and proposed some improvements which convinces me that if I had a second run in this simulation I would finish with less lost orders, a smoother transportation system, an appropriate inventory and ultimately a higher cash balance on day 1460. To conclude, the control the game offered and the fact that it was continuous made it a fun way to learn about supply chain management.
