Excercise UncertainDemand
Excercise UncertainDemand
Excercise UncertainDemand
Newsvendor
Question 1:
Revenue = $35/unit, Cost = $20/unit and salvage value = $15/unit
Demand distribution
Demand X 100 200 300 400 500 600 700 800
Cum Prob
Part (a): What is the optimal lot size?
Part (b): How many units will be sold on average if we buy the optimal lot size? (Use the table below.)
Y (units sold)
Part c): What is the Expected profit?
Part (d): How many units should you buy if there is no salvage value?
Question 2
A factory makes two products. It has capacity to produce a combined total of 100 units per week, in any
combination. Product 1 costs $20 per unit to make, and sells for $30. Demand for this product exceeds
the factory’s capacity. Product 2 costs $10 per unit to make, and sells for $100, but demand is
uncertain. Product 2 is produced in batches once a week. Any units that are not sold by the end of the
week must be thrown out: they have no salvage value. The average demand for Product 2 is 15 units
with a standard deviation of 5. How many units of Product 2, if any, should be produced?
Question 3
Billy’s Bakery bakes fresh bagels each morning. The daily demand for bagels is a random variable with a
distribution estimated from prior experience given by
Number of Bagels
Sold in One Day Probability
0 .05
5 .10
10 .10
15 .20
20 .25
25 .15
30 .10
35 .05
The bagels cost Billy’s 8 cents to make, and they are sold for 35 cents each. Bagels unsold at the
end of the day are purchased by a nearby charity soup kitchen for 3 cents each.
How many bagels should Billy’s bake at the start of each day? (Choose multiple of 5.)
Question 4
a) Sport Obermeyer sources a particular parka from a supplier in China at a (total landed) cost of
$60. The delivery lead time is 8 months, and therefore Sport Obermeyer needs to source it well
before the selling season. The demand for this item over the selling season is normally
distributed with mean 2000 and standard deviation 600. How many units should Sport
Obermeyer buy, if it can sell the item for $120 during the regular selling season and at a 60%
discount at the end of the selling season (assuming Sport Obermeyer wants to maximize
expected profits)
b) Sport Obermeyer considers sourcing additional units of this parka from a domestic supplier,
whenever it runs out of parkas during the selling season. This domestic supplier (whom Sport
Obermeyer already used for the prototype production) has very short lead times, and can
produce the items quick enough to fill additional retailer orders during the selling season.
However, it is more costly and charges $80 per parka. How many units should Sport Obermeyer
buy now from the CHINESE supplier? How many units will Sport Obermeyer buy (in
expectation) from the local supplier?
c) By how much do Sport Obermeyer’s expected profits for this parka increase in expectation, if it
sources from the domestic supplier, whenever it runs out of parkas during the selling season?
Question 5
Part 1: The Park Hyatt Philadelphia at the Bellevue has 118 King/Queen rooms. Hyatt offers a rL= $159
(low fare) discount fare for a mid‐week stay targeting leisure travelers. The regular fare is rH= $225 (high
fare) targeting business travelers. Demand for low fare rooms is abundant, while the demand for high
fare rooms is uncertain and follows a Poisson distribution with mean of 27.3 (see below). Assume most
of the high fare (business) demand occurs only within a few days of the actual stay. How many low fare
rooms will you sell?
Part 2: The forecast for the number of customers that do not show up ( X ) follows a Poisson distribution
with mean 8.5. The cost of denying a room to the customer with a confirmed reservation is $350 in ill‐
will and penalties. How many rooms ( Y ) should be overbooked (sold in excess of capacity)? What is the
chance that Hyatt will turn away customers with reservations?
Poisson distribution with mean 27.3: Poisson distribution with mean 8.5:
Q F (Q ) Q F (Q ) Q F (Q ) Q F (Q ) Q F (Q )
10 0.0001 20 0.0920 30 0.7365 0 0.0002 10 0.7634
11 0.0004 21 0.1314 31 0.7927 1 0.0019 11 0.8487
12 0.0009 22 0.1802 32 0.8406 2 0.0093 12 0.9091
13 0.0019 23 0.2381 33 0.8803 3 0.0301 13 0.9486
14 0.0039 24 0.3040 34 0.9121 4 0.0744 14 0.9726
5 0.1496 15 0.9862
15 0.0077 25 0.3760 35 0.9370
16 0.0140 26 0.4516 36 0.9558 6 0.2562 16 0.9934
7 0.3856 17 0.9970
17 0.0242 27 0.5280 37 0.9697
8 0.5231 18 0.9987
18 0.0396 28 0.6025 38 0.9797
9 0.6530 19 0.9995
19 0.0618 29 0.6726 39 0.9867
Continuous Review Problem
Question 6
For a particular product, the demand per week is normally distributed with N( = 150, = 10). Lead‐
time = 4 weeks. The desired service level is 90%.
(a) What should be the value of reorder point ROP?
(b) What is the safety stock?
Question 7
L = 4 weeks and weekly demand N~[250, 50]. Suppose the company uses a ROP of 1196 units. What
service level is the company providing?
Question 8
A local BP station sells 1,600,000 gallons of regular unleaded gasoline each year and uses a continuous
policy to manage its inventory. It places orders with a fixed order quantity equal to the EOQ quantity,
which is 24,000 gallons. The station’s estimated cost for placing an order is $72; its purchase cost is 91.9
cents a gallon; and its selling price is 106.9 cents a gallon. The costs associated with maintaining a gallon
of gasoline in inventory is annualized at $0.40 / gallon and year. If BP were to run out of gas, it would
lose all sales until its inventory is replenished.
(a) If the following is management’s estimate of lead‐time demand, what is BP’s optimal reorder
point?
Demand (in gallons) Probability
12,000 0.150
14,000 0.600
15,000 0.200
16,000 0.035
18,000 0.010
20,000 0.005
(b) Assume now that both the BP station and its supplier operate 365 days a year. Daily demand is
normally distributed with a standard deviation of 1000 gallons. The lead‐time from supplier to the
station is 3 days. What is BP’s optimal reorder point?
Question 9
Happy Henry’s car dealer sells an imported car called the EX123, and uses a continuous review system.
Happy Henry orders a shipment of new cars on average every 3 months and receives the shipment one
month later. Emergency shipments can be arranged between the regular shipments to re‐supply cars
when inventory falls short of demand. The emergency shipments require two days and buyers are
willing to wait this long for the cars, but without an emergency shipment will generally go elsewhere
before the next regular shipment is due. From experience, it appears that the monthly demand for the
EX123 is normally distributed with a mean of 60 and a variance of 36. The cost of holding an EX123 for
one year is $500. Emergency shipments cost $250 per car over and above normal shipping costs.
a. What ROP should Happy Henry use?
b. Repeat the calculations, assuming that excess demands are back ordered until the next shipment
arrives. Assume a loss‐of‐goodwill cost of $100 for customers having to wait and a cost of $50 per
customer for bookkeeping expenses.
c. Repeat the calculations, assuming that when Happy Henry’s is out of stock of EX123s, the
customer will purchase the car elsewhere. In this case, assume that the cars cost Henry an
average of $10,000 and sell for an average of $13,500. Ignore loss‐of‐goodwill costs for this
calculation.
Question 10
Timbuk2 Inc. manufactures over 100,000 different versions of a messenger bags. Suppose Timbuk2
implements a continuous review model to manage its inventory of a particular plastic buckle. Demand is
approximately normally distributed with a mean of 1000 units per week and a standard deviation of
300. (Each week’s demand is independent of the other weeks.) Since the buckle is sourced abroad, it is
fairly expensive to order the item ($25 per order), and it takes four weeks to receive the shipment. The
annual holding costs for the buckle are $0.10.
(a) How many should Timbuk2 order and when (at which inventory level) should it order, if its
target service level is 99.5%? [how much not on your final]
(b) When should it order, if it uses instead a target fillrate of 99.5%?
(c) What would be the service level provided, if Timbuk2 would use a ROP of 4000 units instead?
(You may assume that Timbuk2 is operating 50 weeks per year.)
Postponement
Question 11
HEC purchases T‐shirts with the HEC logo from a local manufacturer. These T‐shirts are stocked in three
different colors: red, blue, and green. The annual demand for each of these colors is as follows: RED:
2500 units; BLUE: 5000 units; GREEN: 5000 units. Over the past few years, HEC has observed that while
the average annual demand is given by the figures mentioned above, there is a significant variability in
this demand from week to week. HEC estimates that their weekly demand is (approximately) normally
distributed with a standard deviation as follows:
RED: standard deviation of 20 per week; BLUE: standard deviation of 40 per week;
GREEN: standard deviation of 20 per week.
The cost of each of these T‐shirts is 25 Euros. HEC often needs short‐term loans for their purchases, and
they obtain these short‐term loans from a local bank at an interest rate of 15%. HEC also charges itself
an annual warehousing cost that is 10% of the unit cost of items stored, on top of the financial cost of
inventory.
When placing the order, the manufacturer charges HEC 100 Euros per order, to compensate for the cost
of shipment and processing. The lead‐time of replenishment is 2 weeks. As of now, HEC implements
separate continuous review for each of these three items.
(i) Assuming that HEC operates for 50 weeks a year and wants to store enough T‐shirts to provide a
service level of 99%, find the optimal inventory policy (Q and ROP) for each of the T‐shirts.
(ii) What is HEC’s annual inventory cost for these three T‐shirts, if it implements these separate
continuous review policies?
(iii) HEC is considering the purchase of a “Benetton” machine (at the expense of 15,000 Euros) which
can dye the T‐shirts in‐house in one of the three (and more!) colors that are offered for sale. They
order white T‐shirts with the HEC logo, and then dye the T‐shirts in‐house as customers place their
orders. What are the savings induced by the “Benetton” machine? Explain briefly.
Question 12
Sport Obermeyer decided to use black zippers on most of their ski jackets instead of ordering
customized zippers matching the color of the fabric from Japan. How does Sport Obermeyer benefit
from using standard black zippers rather than customized zippers, if they still have to place their order
well ahead of any demand information for their jackets?
Question 13
Home Depot is implementing a new ordering system. Historically, each store ordered from the vendors
separately and orders were sent directly to the stores. With the new system, the stores are required to
order from each vendor on a specific day of the week (e.g., every Monday), all orders are consolidated
and labeled for the individual stores by the vendor, and the items are cross‐docked in Philadelphia.
(i) Which inventory model are the stores now required to use?
(ii) What is the main benefit of the new system over the old one?
Question 14
A. In sectors characterized by uncertain demand and long lead times, such as the apparel
industry, investing in tools and techniques to improve market forecasts is the most effective way
to avoid end‐of‐season sales at a discounted price to get rid of excess inventory.
a) Yes.
b) No.
B. Newsboy losses are a serious concern for many firms in the fashion industry. Which one
between the two companies below is more likely to suffer from these losses? Why?
a) Marks & Spencer (traditional clothing retailer)
b) Zara
Solutions
Newsvendor
Question 1:
Demand distribution
Demand X 100 200 300 400 500 600 700 800
(a): SL = (35 ‐ 20) / (35 – 15) = 0.75; Q = 600
(b): Average units sold = 0.05*100+0.12*200+0.15*300+0.23*400+0.17*500+0.28*600=419
Y (units sold) 100 200 300 400 500 600 600 600
(c): Units bought = 600, average sold = 419, average unsold = 181.
Average profit = 419 * (35 – 20) – 181 * (20 – 15) = $5380
(d): Now salvage value = 0, SL = (35‐20) / (35 – 0) = 0.42856; Q = 400
Question 2
Product 1 has a contribution of $10 per unit. Producing one more unit of Product 2 means we produce
one less unit of Product 1, thereby giving up $10. This represents the additional opportunity cost of
producing product 2, above and beyond the usual costs.
Excess cost =10+10=$20 (production cost + opportunity cost of not selling product 1)]
Shortage cost = 90‐10=$80 (lost margin on product 2 – margin made on product 1);
Critical fractile = 0.8 ; z=0.84; Production quantity: 15+0.84*5 = 19.2 => Produce 20 units of product 2.
Question 3
P = 35 c = 8 v = 3, Cu = 35‐8 = 27 Co = 8‐3 = 5
Critical fractile = 0.84375 => order 25
Question 4
(a) Cu = (120‐60) = 60; Salvage = 0.4 * 120 = 48, Co = (60‐48)=12
Critical fractile =Cu / (Co + Cu) = 60/ (12+60) = 83.33; z = 0.974 (from standard normal table)
Order quantity = 2000 + 0.974 * 600 = 2585
(b) Cu = 80‐60=20; Co unchanged
Critical fractile = Cu / (Co + Cu) = 20 / (12+20) = 62.5; z = 0.319
Order quantity China = 2000 + 0.319*600 = 2192
Exp. order quantity local supplier = expected shortages = L(z) * = 0.2597 * 600 = 155.74
(c) Exp. shortages part (a) = L(z) * = 0.0875 * 600 = 52.5
Exp. sales (a) =exp. demand – exp. shortages = 2000 – 52.5 = 1947.5
Exp. sales part (b) China = 2000 – 155.74 = 1844.26
Exp. leftovers part (a) = batch size – exp. sales = 2585 – 1947.5 = 637.5
Exp. leftovers part (b) = 2192 – 1844.26 = 347.74
Exp. profits (a) = (p exp. sales + v exp. leftovers) – c batch size
= 120*1947.5 + 48*637.5 – 60*2585 = $109,200
Exp. profits (b) = (p exp. sales_C + v exp. leftovers) – c_C batch size + (p – c_l)* exp. sales_l
=120*1844.26 + 48*347.74 – 60*2192 + (120 – 80)*155.74 = $112,712
Increase in exp. profits = $3,512
Question 5:
Part 1:
Cu r r 225 159 66
h l 0.2933
Critical ratio: C C rh 225 225
o u
Hence 24 rooms should be protected for high fare travelers. Similarly, a booking limit of 118‐24 = 94
rooms should be applied to low fare reservations.
Part 2:
Since the demand of high end customers is limited, overbooking results in additional sales of low fare
rooms. Further, since the distribution is given in terms of customers not showing up, underage (and
overage) cost are those of overbooking not enough (too much).
Cu 159
0.3124
Cu Co 350 159
Hence the optimal number of overbooked rooms is Y=7. Hyatt should allow up to 118+7 reservations.
There is about F(6)=25.62% chance that less than 7 customers do not show up, and Hyatt will find itself
turning down travelers with reservations.
Continuous Review Problem
Question 6
(a) From Normal tables: Z0.90 = 1.29
ROP = L* R + ZSL * (L)* = 4 * 150 + 1.29 * (4)* 10 = 625.8 ~ 626
(b) Safety Stock = ROP ‐ L* R = 626 – 4*150 = 26
Question 7
ROP = mean * L + z* std. dev. * SQRT(L) = 4 * 250 + z * 50 * SQRT (4) = 1196
z = (1196 ‐ 1000)/100 = 1.96. From standard normal table: Service Level = 97.5%
Question 8
(a) D= 1,600,000 gallons/year S= $72 per order h*C= $0.40 per gallon‐year
2 DS
EOQ = Q* = 24,000 gallons [given in revised question]
h *C
Q
Cu= (106.9 ‐ 91.9) cents = $0.15 per gallon Co = h * C * = 0.006
R
CO 0.15
Service level = 0.962
C O CU 0.156
Demand (in gallons) Probability Cum. Prob.
12,000 0.150 0.150
14,000 0.600 0.750
15,000 0.200 0.950
16,000 0.035 0.985
18,000 0.010 0.995
20,000 0.005 1.000
To satisfy Pr(lead time Demand ROP) service level ROP=16,000
Q = 24,000 ROP = 16,000
(b) mean lead‐time demand = (160000/365)*3 = 13150
std dev of lead‐time demand =1000*sqrt(3) = 1732
service level = 0.962 From table, we need to find the z value which corresponds to the 96.2
percentile of the standard Normal Distribution; a table look‐up gives z = 1.78
ROP = mean lead‐time demand + (z * std dev. of lead‐time demand)
= 13150 + (1.78 *1732) = 16232.96
Q = 24,000 ROP= 16,232.96
Question 9
500
a) Cu = 250 Co = = 125
4
250
service level = = 0.6667 from table, Pr (z 0.6667) implies z = 0.44
375
ROP= 60 + 0.44 * 36 62.64 63 => order 63 – on‐hand
500
(b) Cu = 100 + 50 = 150 Co = 125
4
150
service level 0.5455 Pr( z 0.5455) implies z 0.12
275
ROP = 60 + 0.12 * 6 = 60.72 61 => order 61 – on‐hand
(c) Cu = 13,500 ‐ 10,000 = 3,500; Co = 125
3500
service level 0.9655 Pr(z 0.9655) implies z = 1.82
3500 125
Order‐up‐to level = 60 + 1.82 * 6 = 70.9 71 => order 71 – on‐hand
Question 10
(a) How much to order: EOQ = Sqrt(2*1000*50*25/0.1)=5000
ROP = 1000 * 4 + 2.576 * 300 * Sqrt(4)=5546 (round up!)
(b) Expected shortages per cycle= 0.5%*1000*4 = L(z)*300*2 => L(z) = 0.03333 => z=1.44 => ROP =
4*1000+1.44*300*2=4864
(c) New SL: z=(ROP – 1000*4)/(300*2) = 0; hence the SL=50%
Postponement
Question 11
Part (i) QRed=EOQ=Sqrt(2*2500*100/(25*0.25))=283, QBlue=QGreen=400
ROPRed= 2500/50*2+z*20*Sqrt(2)=166; ROPBlue= 332; ROPGreen=266
Part (ii) Avg Inventory Cost =(Q/2 + safety stock)*H
Inv. CostRed = (283/2+66)* 25*0.25=1296.87; Inv. CostBlue =2075; Inv. CostGreen =1662.5
Total avg inventory cost = 5034.37
Part (iii) You can expect two types of savings over the current system. One comes from the fact that you
are placing now joined orders for all sweaters, so HEC will incur the ordering costs only once for all
sweaters. This should also allow you to lower the (total) order quantity per shipment (i.e.,
Qwhite<QRed+QBlue+QGreen), thereby reducing the cost for cycle inventory. The second savings are due to
risk pooling. Assuming less than perfect correlation in demand, the total demand across all colors is less
variable than the demand for the individual sweaters. Therefore, HEC would need to hold a lower safety
stock to provide the same customer service (i.e., safety stockwhite<safety stockRed+safety stockBlue+safety
stockGreen), thereby reducing the cost for safety stock inventory. [Alternatively, HEC could keep the
safety stock the same and provide a higher customer service.]
Question 12
Risk pooling => lower safety stock required for the standard zipper rather than customized zippers, and
less shortages. (Possibly also lower unit costs and possible lower transportation costs if it makes local
sourcing possible.)
Question 13
(i) Periodic Review
(ii) Lower transportation cost due to joined shipments
Question 14
A) Answer: b) It is typically extremely difficult and costly to improve forecasts beyond a given level
of accuracy. A better way of dealing with newsboy losses would be to reduce lead‐time (which
reduces uncertainty and implicitly makes forecasting easier and more accurate) or to supply a
continuously renewed set of small batches that never saturate the market, as Zara does.
B) Marks & Spencer. Newsboy losses are a serious concern for companies that – because of their
long lead time ‐ need to forecast demand ahead of time. This is the case for Marks & Spencer,
which have on average much longer lead times than Zara. Alternatively one could argue that
Marks & Spencer operate in a much more stable market than Zara (traditional UK customers as
opposed to young urban shoppers). Hence – in theory – it would be easier for them to predict
demand and to hedge against newsboy losses. This argument is less strong than the first one,
though.