ISyE 3104: Introduction to Supply Chain Modeling: Manufacturing and Warehousing

Instructor: Spyros Reveliotis

Spring 2002
Homework #5
Due Date: Friday, 4/5/02
Problem set:
1. From your textbook: Problems 5.1, 5.2, 5.3, 5.4, 5.10 and 5.11 (for 5.11, just provide the formulation
modeling the new situation).
5.1 Advantages of Aggregating Items
Less data is required for building and maintaining models.
Less computation required to solve model (larger systems can be modeled).
Forecasts of aggregate demand are usually more accurate than those of individual items.
Allows short term scheduling to react to current inventory levels when planning batch sizes
and order release times.
5.2 Standard outputs of an aggregate production plan
Production quantity for each product family in each period.
Planned amount of each key resource to be used over the planned horizon.
Inventories and backlogs to be carried during the different periods.
Differences from the outputs of the disaggregate production plan:
Aggregate plan avoid solving unnecessary detailed data
5.3 Major tradeoffs examined in aggregate production planning models are:
The major tradeoffs are among inventory, workforce/equipment changes, workload
changes/overtime, and product shortage. An example would be:
 Inventory holding cost VS Shortage cost
 Cost to modify capacity VS holding cost and shortage cost
5.4 Assume that the machine hours required for each product type are known.
Decision Variables:
Xit = amount of product i made in period t;
Iit = inventory level of product i at the end of period t;
Wjt = workforce level of worker type j in period t;
Hjt (Fjt) = Number of workers of type j Hired (Fired) effective period t.
t t
Note that inventory variables can be eliminated from the model by noting Iit  Ii 0   Xir   Dir and
r 1 r 1

then replacing all Iit variables.

a. Xit variables: (1500 products) x (18 months) x (4 weeks/month) = 108,000.
Wjt, Hjt , Fjt variables: (12 labor classes) x (18 months) x (4 wks/mth) x (3 vars) = 2,592.
Thus, approximately 110,000 variables.
b. For the aggregated model we have 4 products, 1 labor class, 36 periods. This requires 144 Xit variables
and 36 variables each for workforce level, hiring, and firing.
5.10 Let Xit = production of product family i in period t;
Sijt = shipments of product family i from the plant to warehouse j in period t;
Iijt = inventory of i at location j in period t (j = 0 for plant, j > 0 indicates warehouse);
Wt = number of workers employed in period t;
Ht (Ft) = number of workers hired (fired) in period t;
Ot = overtime hours scheduled in period t.
3  2  2 
Minimize Cost =   20(120)Wt  30Ot  400 H t  500 Ft   
i 1 
0.8 I i 0t  0.6  I ijt  

t 1  j 1 
subject to:
5 X1t  6 X 2 t  120Wt  Ot  60 H t , t , Labor Availability

Ii 0 t  Ii 0,t 1  X it  Si1t  Si 2 t ,  i, t, Plant Inventory Balance

Iijt  Iij ,t 1  Sijt  Dijt  i, j>0, t, Warehouse Inventory Balance

Wt  0.99Wt 1  Ht  Ft , t Workforce Balance, W0 = 275

I1 jt  I2 jt  3000, j , t Warehouse Storage Capacity

Ot  20Wt  20 H t t Overtime Limit (Assumes no new workers)

Ii 0 t  0, Iijt  500, X it  0
5.11 Assumption: Workers may not be hired or fired. (There will be no Ht and Ft.)

3 2  2 
Minimize Cost =   20(120)Wt  30Ot    0.8 I i 0t  0.6 I ijt  
 i 1 

t 1  j 1 
subject to:

5 X 1t  6 X 2t  120Wt  Ot , t , Labor Availability

Ii 0 t  Ii 0,t 1  X it  Si1t  Si 2 t ,  i, t, Plant Inventory Balance

Iijt  Iij ,t 1  Sijt  Dijt  i, j>0, t, Warehouse Inventory Balance

Wt  0.99Wt 1 , t Workforce Balance, W0 = 275

I1 jt  I2 jt  3000, j , t Warehouse Storage Capacity

Ot  20Wt t Overtime Limit

Ii 0 t  0, Iijt  500, X it  0
2. A local firm manufactures children's toys. The projected demand over the next four months for one

Month Workdays Forecasted Demand

July 23 3,825
August 16 7,245
September 20 2,770
October 22 4,440
particular family of toy robots is as follows:
Assume that a normal workday is 8 hours. Hiring costs are $350 per worker and firing costs (including
severance pay) are $850 per worker. Holding costs are $4.00 per aggregate unit held per month.
Assume that it requires an average of 1 hour and 40 minutes for one worker to assemble one toy.
Shortages are not permitted. Assume that the ending inventory for June was 600 of these toys and the
manager wishes to have at least 800 units on hand at the end of October. Assume that the current
workforce level is 35 workers.
a. Find the minimum constant workforce plan for the four months and the total hiring, firing and
holding costs of that plan.

Solution:
Since there is a 600 units ending inventory in June, the demand for July will be reduced to be
3825– 600 = 3225 units. By the same way, the demand for October will be increased to be 4440 + 800
= 5240 units. This is based on First In First Out policy.

Month Demand Days Units/workr/mo. Cumulative Cumulative Worker Req.

Units/workr Demand
July 3225 23 110.4 110.4 3225 30
August 7245 16 76.8 187.2 10470 56
Sept 2770 20 96 283.2 13240 47
Oct 5240 22 105.6 388.8 18480 48

Note:
1. To calculate the number of units produced per worker for a particular month (column 4), we
first calculate the number of units produced per day, which is equal to (8 hr)/(1 hr 40 min/ unit) =
4.8 units/day. Thus, the number of units produced per month is Days available * 4.8. For example,
in July, unit/worker/mo. = 23 days * 4.8 units/day – 110.4 units/worker/mo.

2. To calculate the number of workers required for meeting the demand without any stockouts
(column 7), for every month, we divide the cumulative demand up to that month (column 6) by the
cumulative number of units produced by a worker (column 5) over the corresponding time
interval. Then, we know that for any given month, the cumulative demand up to that month will be
met if we employ the number of workers indicated in the corresponding cell, and have them work
full-time until that month, possibly building anticipatory inventories in certain months, to be
consumed in the subsequent months. Obviously, if we want to experience no stockouts over the
entire planning horizon, we must employ the maximum number of workers appearing in the
aforementioned column (i.e., 56). These workers will need to work full-time for the months of
July and August, but they will be under-utilized in the sub-sequent months of September and
October.

Table below calculates the production based on 56 workers:

Month Cumulative Production Cumulative Ending Inv.

Demand Production
July 3225 6182 6182 2957
August 10470 4288 10470 0
Sept 13240 2770 13240 0
Oct 18480 5240 18480 800
 The total inventory held is 2,957 + 0 + 0 + 800 = 3757 units, which results in an inventory holding cost
of (3757)(4) = $15,028. The cost of hiring is (56-35)(350) = $7,350, giving a total cost of $15,028 +
$7,350 = $22,378.

b. Determine the plan that corresponds to demand chasing, i.e., the plan that changes the workforce
level each month to most closely match the demand, and compute the cost(s) of this plan.

Month Demand Days Unit/worker/mo. Worker Req./mo.

July 3225 23 110.4 30
August 7245 16 76.8 95
Sept 2770 20 96 29
Oct 5240 22 105.6 50

Month Cumulative Production Cumulative Ending Inv.

Demand Production
July 3225 3225 3312 0
August 10470 7245 10608 0
Sept 13240 2770 13392 0
Oct 18480 5240 18672 800

The total number of workers fired is 5+66 = 71 at a cost of (5+66)(850) = $60,350 and the total
hired is 65+21 = 86 at a cost of (86)(350) = $30,100. inventory holding cost is (800)(4) = $3,200. The total
cost of this plan is $60,350 + $30,100 + $3,200 = $93,650.

Note: Since the problem does not give us any information about the labor rates, we cannot compute any
(regular) labor costs (which is also part of the cost structure in the considered problem context).
Furthermore, these labor costs include also an "undertime" cost, since in certain months there is a certain
amount of time paid to the workers for doing nothing - this is the results of rounding to the next intege /
employing the "ceiling" function, in the computation of the worker requirements.

3. Computing a Master Production Schedule for the McGuinnes & Co. Microbrewery.
In this problem you are invited to use the tabular (spreadsheet-based) approach to develop a 6-month (26
weeks) MPS for the McGuinness microbrewery case study presented in class. A complete description of
the case study and the spreadsheet discussed in class that can support your calculations, can be downloaded
from the course Web-site (http://www.isye.gatech.edu/~spyros/courses/IE3104/course_materials.html). A
detailed description of the faced situation is as follows:
a. At the beginning of 2001, McGuinness & Co. had in its inventory the following quantities from each
of the five products:
Product Avail. Inventory (in cases)
Pale Ale 800
Stout 400
Winter Ale 750
Summer Brew 0
Octoberfest 0
Furthermore, at that point, the company was brewing a full fermentor for each of the first two products,
with these production lots requiring another week of fermentation for the pale ale (i.e., this lot would
be available at the beginning of week 2), and two more weeks of fermentation for the stout (i.e., this lot
would be available at the beginning of week 3).
Finally, using the forecasting techniques discussed in class, and the information available in his order
records, Mr. McGuinness was contemplating that the demand for each of the company five products
over the next six months (26 weeks), would evolve as follows:
Product W1 W2 W3 W4 W5 W6 W7 W8 W9 W10 W11 W12 W13
Pale Ale 350 340 300 300 300 300 300 300 300 300 300 300 300
 Stout 170 170 160 160 165 165 165 165 175 175 175 175 190
Winter Ale 320 330 310 310 265 265 265 265 180 180 180 180 150
Summer Brew 40
Octoberfest
Product W14 W15 W16 W17 W18 W19 W20 W21 W22 W23 W24 W25 W26
Pale Ale 300 300 300 300 300 300 300 300 300 300 300 300 300
Stout 190 190 190 200 200 200 200 215 215 215 215 225 225
Winter Ale 150 150 150 50
Summer Brew 40 40 40 100 100 100 100 170 170 170 170 225 225
Octoberfest

Using the above data, and the information in Word document describing the McGuinness & Co.
Microbrewery case study regarding (i) the production (fermentation) lead times for each brew, (ii) the
production capacity of the microbrewery, and (iii) the company policy regarding the maximal allowed
shelf-life, develop a production schedule for the next 26 weeks. In the preparation of this schedule
you should also consider that, under the current operational conditions, the company produces the
various products only in lots of a full or half fermentor (this was, indeed, the situation for the
senior design project!)

In your work, you can utilize the spreadsheet presented in class (and available at the course Web-site).
Remember, that the green cells in this spreadsheet correspond to the input data required for the
problem definition, the orange/brown cells introduce your scheduling decisions (i.e., when to initiate
production for each product, and how much to produce in each lot) in the overall computation, while
the remaining white cells implement all the additional computations needed in order to evaluate the
feasibility and quality / performance of your schedule. Also, as indicated in class, the overall
spreadsheet computation is organized in a number of segments, with the top segment assessing the
feasibility of any tentative schedule w.r.t. the available production capacity (i.e., fermentor capacity
and availability), while the remaining segments compute the projected inventory position for each
product, based on the current product availability (i.e., initial inventory position and scheduled
receipts), its demand profile, and the contemplated scheduled releases (i.e., the (tentative) production
plan w.r.t. this product).

Notice that there is a possibility that you will find out that the satisfaction of the entire (projected)
demand with the current production resources and product availability is impossible. In that case, and
considering the fact that the company experiences intense growth, the best strategy for Mr.
McGuinness would be to expand its production capacity by buying and installing another (or possibly
more) fermentor(s). However, in your work consider that, due to financial constraints, the
company cannot afford the purchase of another fermentor in the immediate future, and discuss
how Mr. McGuinness should deal with any arising schedule infeasibilities.
At the end of this part of your work, please, turn in (i) the proposed production schedule, (ii) the
resulting spreadsheet, and (iii) a supporting document explaining and justifying your
proposition.
Solution:
For the specified problem data, there is no feasible production schedule that can satisfy the entire product
demand over the considered planning horizon. Hence, assuming that no additional production capacity
(fermentors) can be installed during the considered time interval, we must consider how to accommodate
the schedule infeasibility, by deciding which part of the demand should be left uncovered.

This decision should take into consideration the product phases w.r.t. (i) their entire life-cycles, as well as
(ii) their seasonal cycles, and their implications for the company’s marketing and distribution policies. One
way to reason about this problem is as follows:

1. The pale ale seems to be the major company product (i.e., the most well-established and with the most
extensive circulation), having reached its mature phase. Hence, the company should maintain a stable
and responsive production for this product. Furthermore, the current product expansion to a new
 market (Northeast) implies that the company must be careful with its new undertaken obligations
(deals and/or contracts with the new distributors and customers). Based on these considerations, it
seems that the demand for pale ale should be met on its entirety.
2. The stout is a brew that is in its growth phase, and it is developing to the second major product for the
company. Therefore, it is to the company’s advantage to promote aggressively this product. Allowing
for planned shortages does not seem to support the company interests.
3. Similarly, the summer brew is a product in its growth phase. Furthermore, the product demand in the
considered planning horizon corresponds to the “opening period” for the product. As a result, the
company should try to meet the entire expected demand for this product, as well, since in this way it
would promote and secure the product position in the market.
4. Finally, the winter ale is a product that is rather well-established; the product has been around for a
while, and its annual demand presents some stability. At the same time, this is a seasonal product for
the company, and the estimated product demand for the considered planning horizon – especially the
last few weeks - corresponds to the closing phase of the product seasonal cycle. The previous two
remarks imply that the company might be able to “take a hit” w.r.t. this product, by terminating the
product distribution a little earlier than planned for this year, without this fact affecting significantly
the product marketability and the company image.

To initialize the excel spreadsheet, we will follow the following steps:

Step 1: Specify the number of fermentors, amount produced per fermentors, and product shelf life.

Step 2: Put in the fermentation time and the initial inventory level, i.e. 2 week fermentation time and initial
inventory level of 800 for pale ale.

Step 3: Key in the given demand of each products

Step 4: Introduce the scheduled receipts, taking into consideration that the company is currently brewing
one fermentor of pale ale and one fermentor of stout, to be ready by weeks 2 and 3 respectively.

Now we are ready to experiment with the spreadsheet in order to determine the scheduled releases that will
"minimize" the implications of the experienced shortages.

A schedule developed according to the logic outlined above is as follows:

b. Suppose that Mr. McGuinness hired the services of an accountant who, after some thoughtful analysis
of the company operations, informed him that his operational cost breakdown is as follows:
i. Every initiation of a new production lot costs $s1 if the utilized fermentor had been used
previously for the production of the same type of beer, and $s2 dollars otherwise.
ii. Carrying one case of inventory from on week to the next costs $hi, i=1,…5, depending on the type
of beer.
iii. Backordering a case of beer for one week costs $bi, i=1,…,5, depending on the type of product
being backordered.
iv. Finally, the variable production cost of producing one case of beer is equal to $c.

Develop a formula that estimates the total cost of the production schedule developed in part (a). (As
you can see, once this formula is implemented in the provided spreadsheet, it can allow the
development of MPS's that are evaluated and "optimized" according to the more "standard"
considerations of inventory control theory.)

Solution:

Total cost = Production cost + Holding cost + Backordering cost + Setup cost

5 T
Production cost =  cX
i 1 t 1
it where Xit is the number of unit of product i produced at time t

5 T
Holding cost =  h I
i 1 t 1

i it where Iit+ is the on-hand inventory of product i at time t

5 T
Backordering cost =  b I
i 1 t 1

i it where Iit- is the backorders of product i at time t

5 T 5 T
Setup cost =  s1Fit Pit   s2 Max( Fit , Fi,t 1 )(1  Pit )
i 1 t 1 i 1 t 1
Where Fit is the number of fermentors occupied by product i at time t
Pit is equal to 1 if Fit = Fi, t-1; 0 otherwise

c. An additional concern that arises in the development of feasible production schedules is the
accommodation of any preventive maintenance of the production equipment that might reduce the
nominal production capacity over certain periods of the planning horizon. Suppose the McGuinness &
Co. Micro brewery runs a preventive maintenance program that necessitates the removal of each
fermentor for certain predetermined time intervals over the considered planning horizon (measured in
weeks - typically in the order of one to three weeks since these fermentors might have to be taken to
the facility of the sub-contractor supporting the maintenance program). Discuss how you would
modify the provided Excel spreadsheet in order to incorporate the maintenance effects in your
calculations.

Solution:

There are two possible ways to accommodate the preventive maintenance.

1. We can replace the “number of fermentors” cell on the very top of the spreadsheet, by an entire
input row that will allow us to state the fermentor availability on a period-by-period basis.
2. We can add another imaginary (fictitious) product that would occupy the fermentors during the
periods of scheduled maintenance. For this example, we can assume and use the Octoberfest
Lager as a maintenance requiring product. You need to specify the fermentation time to be 1
week. So, if you have a three-week maintenance, you will put schedule release on all three weeks.
Extra credit (50%)
Provide a mathematical (mixed integer) programming formulation for production scheduling problem as it
arises in the McGuinness case study presented in class. In your formulation, consider that backorders are
not allowed, and the optimized objective is the minimization of the inventory holding cost (i.e., set up cost
is not an issue - the main value of this formulation is that it would provide a definite answer on whether
there exists a feasible production schedule or not).

Please, refer to the following (pdf) document:

http://www.isye.gatech.edu/~spyros/courses/IE3104/mcguinness-prod-plan.pdf