7.

3 Solutions to the Problem Set

Problem 7.1 [a] Given quantities: mean weekly demand = 400; standard deviation of weekly demand = 125; replenishment lead time = 1 week and reorder point (ROP) = 500 units. We compute the average demand during leadtime to be 400 units. Thus the safety stock, Isafety = 100 units; to find the service level provided, we need to find the area under the normal curve to the left of the reorder point (ROP) = 500. Let the demand during lead time be LTD. The cycle service level Prob( LTD ROP) = Prob( LTD R + Isafety) = 0.7881. [b] So the cycle service level is 78.81%. The standard deviation of lead time demand, LTD = 125 units. For each service level the z-value can be read from the standard normal table. The safety inventory Isafety = z x LTD . Finally, ROP = 400 + Isafety. Cycle Service Level z= Isafety = ROP = 80% 0.842 105 505 90% 1.282 160 560 95% 1.645 205 605 99% 2.326 290 690

Problem 7.2

[a]

Average weekly demand (R) = 1000 Standard deviation of weekly demand (R) = 150. Lead time (L) = 4 weeks. Standard deviation of demand during lead time (LTD ) = L R = 300. Current reorder point (ROP) = 4,200. Average demand during lead time (LTD) = L x R = 4,000. Current level of safety stock (Isafety)= 200. Current order quantity (Q) = 20,000 Average inventory (I) = Isafety + Q/2 = 200 + (20,000/2) = 10,200. Average time in store (T) = I/R = 10,200/1,000 = 10.2 weeks. Annual ordering cost = S x R/Q = $100 x 2.5 = $250. Annual holding cost = H x I = $0.25 x 10,200 = $2,550.
61

[b]

We use the EOQ formula to determine the optimal order quantity. H = $1 * 25%/year = $.25/year R = 1,000 /week = 50,000/year
S = $100. Thus, the economic order quantity is Q=
2 RS 2 50,000 100 = 6,325 units. = H 0.25

To determine the safety inventory, Is, for a 95% level of service, we first observe that the z-value = 1.65. Then Isafety = z x LTD = 1.65 x 300 = 495. Average inventory (I) = Isafety + Q/2 = 495 + (6,325/2) = 3,657.5. Average time in store (T)= I/R = 3.6575 weeks. [c] If lead time (L) reduces to 1 week, then standard deviation of demand during lead time (LTD) = 150. Safety stock for 95% level of service = 1.65 x 150 = 247.5. Average inventory = 247.5 + (6,325/2) = 3,410. Average time in store = 3.41 weeks.

Problem 7.3
7.3 The Home and Garden (HG) chain of superstores imports decorative planters from Italy. Weekly demand for planters averages 1,500 with a standard deviation of 800. Each planter costs $10. HG incurs a holding cost of 25% per year to carry inventory. HG has an opportunity to set up a superstore in the Phoenix region. Each order shipped from Italy incurs a fixed transportation and delivery cost of $10,000. Consider 52 weeks in the year. The optimal order quantity of planters for HG is (i) (ii) (iii) (iv) About 24,980 About 3,464 About 78,994 None of the above
* Q =

(a)

2 RS 2 1500 52 10000 = = 24,980 H 2.5

(b)

If the delivery lead time from Italy is 4 weeks and HG wants to provide its customers a cycle service level of 90%, how much safety stock should it carry? (i) (ii) (iii) (iv) About 1,025 planters About 2,050 planters About 4,100 planters None of the above

Safety stock = NORMSINV(.9)*sqrt(4)*800 = 2050

(c)

Fastship is a new shipping company that promises to reduce the delivery lead time for planters from 4 to 1 week using a faster ship and expedited customs clearance. Using fast ship will add $0.2 to the cost of each planter compared to the current approach. Should HG go with Fastship? Why? Quantify the impact of the change. Additional transportation cost per year = 1500*52*.2 = $15,600 Savings in holding cost = NORMSINV(.9)*800*(sqrt(4)-sqrt(1)) = $2562.5 Thus Fastship should not be used.

Problem 7.4 First, is this an EOQ problem? Well, notice that the question dictates that we do a run every two years. That would mean, in a deterministic EOQ setting, that Q must equal two years of mean demand, i.e., 32000. Hence, this question does not give us the freedom to change when we do a run (which is what EOQ is all about). Thus, the question is whether 32000 is the best quantity we can print every tow years? This thus asks about what the appropriate safety stock (or service level) should be. We know that this is answered via newsvendor logic. Answer these two questions: 1. What is my underage cost (cost of not having enough)? I.e., if I were to stock one more unit, how much could I make? Every catalog fetches sales of $35.00 and costs $5.00 to produce. Thus, the net marginal benefit of each additional unit (MB), or the underage cost, is p c = $35 - $5= $30. 2. What is my overage cost? I.e., if I had stocked one less unit, how much could I have
saved? The net marginal cost of stocking an additional unit (MC) = c v = $5 0 = $5.

Now, we can figure out the optimal service level (or critical fractile): SL = 30/(30+5) = 0.857. The last step is to convert the SL into a printing quantity. Recall that total average demand for 2 years (R) = 32,000 with a standard deviation of 5656.86. The optimal printing quantity, Q* is determined such that MB 30 = = 0.857 . Prob(R Q*) = ( MB + MC ) 30 + 5 The optimal order quantity Q* = R + z where z is read off from the standard Normal tables such that area to the left of z is 0.857. That is, z = 1.07. This gives Q* = 38,053 catalogs. It can be verified that the optimal expected profit (when using Q* = 38053) is larger than $25,000, the fixed cost of producing the catalog.

Problem 7.5
The revenue per crate, p = $120.00, variable cost, c = $18.00, and salvage value, v = $2.00. The marginal benefit of stocking an additional crate (MB) = p c = $120 - $18 = $102. The marginal cost of stocking an additional unit (MC) = c v = $18 + $2 = $20. Then MB/(MB+MC) = 102/(102+20) = 0.836. The probability density of demand and its cumulative probability is listed below.

Demand Frequency Prob. Cumulative Prob.

0 0 0 0

1 0 0 0

2 0 0 0

3 1 0.02 0.02

4 3 0.06 0.08

5 2 0.04 0.12

6 5 0.1 0.21

7 1 0.02 0.23

8 6 0.12 0.35

9 7 0.13 0.48

10 6 0.12 0.6

11 8 0.15 0.75

12 5 0.1 0.85

13 4 0.08 0.92

14 1 0.02 0.94

15 3 0.06 1

The optimal order quantity is the smallest number of crates such that cumulative probability is at least 0.836. From the table this gives the number of crates to be 12. Problem 7.6
The residents of Bucktown, Illinois, place their trash at the curb each Wednesday morning to be picked up by municipal crews. Experience shows that the total amount trash has a normal Trashof put out put outWe save $650 ondistribution our additional ton that noassigned longer R> Q with a mean of 35 tons and a standard deviation of 9 tons. Crews of full-time city employees to must be picked up by the trash collection collect trash. Each crew can collect 5 tons of trash per working day. The city has plenty outside contractor Say wetrash have collection. The marginal cost of operating of trucks of the kind used for one trash collection crew planned to for one working day, including both personnel-related costs and truck-related costs is reckoned at $625. pick up Q Whatever trash remainstons at the end of the work day must be collected that evening by an outside contractor R who charges $650 per ton.
Trash put out Assign to pick up Our additional ton didn't R<Q one additional ton make a difference. How many crews should the city assign to trash collection? For simplicity, you may treat the MC = $625/5ton = We spent extra $125 number of crews as a continuous variable. For example, 4.1 crews would be a perfectly acceptable $125/ton

answer.

One solution approach: Hence, MC = $125/ton = expected MR = $650 Prob(R>Q) Prob(R>Q) = 125/650 or SL = 525/650 = 80.77% z = .87 Q = 35tons + .87*9tons = 42.8 tons = 8.56 crews. Another approach to get the critical fractile probability SL Here we are stocking up on local trash collection capacity. Cost of overstocking by 1 ton =MC = 625/5 = $125 Cost of understocking by 1 ton = additional cost of using outside trash pick up = MB = $650-$125 = $525 SL = Prob(R Q) = MB /( MB + MC) = 525/(125 + 525) = 0.8077 appropriate # of crews = 8.56 crews

Problem 7.7 We are concerned about the overbooking problem; that is, how many seats to overbook. The demand randomness in the demand arises from uncertain cancellations, which is uniformly distributed between 0 and 20. The marginal cost of a flying an empty seat is the lost revenue; thus MB = $600. The marginal cost of an overbooked seat is the cost incurred when the passenger shows up but cannot be accommodated as the flight is full. The total marginal cost of overbooking is $250+$600 = $850. Thus optimal service level is MB/(MB+MC) = 0.41. The optimal overbooking quantity is determined by Prob(R Q) = MB/(MB+MC) = 0.41. For uniform distribution between 0 and 20, Prob(R Q)=Q/20. Thus Q = 20 * 0.41 = 8.2 seats. Rounding down to integer number of seats, we determine that the optimal overbooking level is 8 seats.

Problem 7.8 [a] To compute the optimal order quantity at each store we use the EOQ formula. Assume 50 sales weeks/year. H = $10 * 25%/year = $2.5/year R = 10,000 /week = 500,000/year S = $1000. Thus,
Q = EOQ =
2 RS 2 500,000 1000 = 20,000 units. = H 2.5

The replenishment lead time (L) = 1week. Standard deviation of demand during lead time at each store () = 2,000. Safety stock at each store for 95% level of service (Is) = 1.65 x 2,000 = 3,300. Reorder point (ROP)= + Is = 10,000 + 3,300 = 13,300. Average inventory across four stores (Id) = 4 x (Is + Q/2) = (3,300+(20,000/2)) = 53,200. Annual order cost for all four stores = 4 x S x R/Q = 4 x 1,000 x25= $100,000. Annual holding cost for all four stores = H x Id = $133,000. Average time unit spends in store (T) = Id / 4 x R = 53,200/40,000 = 1.33 weeks.

[b] To compute the optimal order quantity at centralized store observe that this store faces a cumulative average weekly demand = 4 x 10,000 = 40,000. This gives an annual demand of 2,000,000 units.
Q = EOQ =
2 RS 2 2,000,000 1000 = 40,000 units. = H 2.5

Standard deviation of demand during lead time at central store () = 4 2000 = 4,000. Safety stock at central store for 95% level of service = 1.65 x 4,000 = 6,600. Reorder point (ROP) = 40,000 + 6,600 = 46,600. Average inventory in central store (Ic) = (6,600+(40,000/2)) = 26,600. Annual order cost for central store = S x R/Q = $1,000 x 50 = $50,000. Annual holding cost for central store = H x Ic = $66,500. Average time unit spends in store (T) = Ic / 4 x R = 26,600/40,000 = 0.67 weeks. Problem 7.9
Hi-Tek is a retailer of computer equipment in the Chicagoland region with four retail outlets. Currently each outlet manages its ordering independently. Demand at each retail outlet averages 4,000 units per week. Each unit costs $200 and Hi-Tek has a holding cost of 20%. The fixed cost of each order (administrative + transportation) is $900. Assume 50 weeks in a year. (a) Given that each outlet orders independently and gets its own delivery, the optimal order size at each outlet is (i) (ii) (iii) (iv) (v) (b) sold? 424 3,000 = sqrt(2RS/H) = sqrt(2*4000*50*900/(.20*200)) 6,000 42,426 None of the above On average, how long (in weeks) does each unit spend in the Hi-Tek system before being

T = I / R = (Q/2) / R = 1,500 / 4000 weeks = 3/8 weeks = .375 weeks

(c) Hi-Tek is thinking of centralizing purchasing (for all four outlets). In this setting, Hi-Tek will place a single order (for all outlets) with the supplier. The supplier will deliver the order on a common truck to a transit point. Since individual requirements are identical across outlets, the total order is split equally and shipped to the retailers from this transit point. This entire operation has increased the fixed cost of placing an order to $1,800. If Hi-Tek manages ordering optimally in the new setting, average inventory in the Hi-Tek system (across all four outlets) can be expected to (i) (ii) (iii) Increase Decrease Remain unchanged

8.3 Solutions to the Problem Set

Problem 8.1
Increasing the number of WATS lines from 12 to 13 will a. Decrease the proportion of customers who get a busy signal. Callers calling when there are 12 people in the system will no longer get a busy signal. b. Increase the average flow time experienced by customers since some customers served will have waited for 12 customers ahead of them to be served. In the old system (with 12 lines) no customer waited for more than 11 customers ahead of them to be served. c. Increase the server utilization since the same number of servers will serve more customers (since fewer are blocked).

Problem 8.2
Merging the two departments will a. Decrease the proportion of callers getting a busy signal. b. Decrease the average flow time experienced by the callers.

Problem 8.3
John Doe has promised a total service time of 20 minutes. The processing time at the oven is 15 minutes with 2 minutes for preparation. This leaves at most 3 minutes in waiting time before the pizza has to be given away for free. The average demand is anticipated to be 20 pizzas per hour. Since John plans to get 5 ovens (each with a capacity of 60/15 = 4 pizzas per hour) there is no safety capacity. Any variability in arrival rate (or service time) will lead to a build up of queues and waiting time in excess of 3 minutes, resulting in John giving up many free pizzas. Thus the potential partner should either get John to buy more ovens or get out of the business, or stop giving the service guarantee.

Problem 8.4 (M. M. Sprout)

a. We are given: Average arrival rate Ri = 1/4 per minute, Average unit capacity 1/Tp = 1/3 per minute, Number of servers c = 1. Using the PERFORMANCE.xls spreadsheet, we get Server utilization = Ri / Rp = 0.75, Average waiting time Ti = 9 mins, Average number of customers on hold Ii = 2.25, Average number of customers in the system I = 3. The costs include the CSR wages and the cost of waiting (line charge + waiting cost for customers). We have Hourly wages of CSR = $20 / hour, Line charge = $5 / hour (for all lines used), Customer waiting cost = Average number on hold 60 $2 = 2.25 120 = $270. Hence, the total hourly cost = $20 + $5 + $270 = $295/hour.

b. With only four lines and one CSR, we have Average arrival rate Ri = 1/4 per minute, Average unit capacity 1/Tp = 1/3 per minute, Number of servers c = 1, Maximum buffer size K = 3. Using the spreadsheet Performance.xls we get Average waiting time Ti = 3.45 mins, Average # of customers on hold Ii = 0.77, Average number of customers in system I = 1.44, Probability of blocking = 0.104. In this case the costs incurred are the CSR wages, the cost of waiting (line charge + waiting cost for customers) and the lost business because of blocked calls. We have Hourly wages of CSR = $20 / hour, Line charge = $5 / hour, Customer waiting cost = Average number on hold 60 $2 = .77 120 = $92.4, Cost of blocking = Calls blocked per hour $100 = Probability of blocking Average arrival rate $100 = 0.104 15 $100 = $156. This implies that Total hourly cost = $20 + $5 + $92.4 + $156 = $273.4.

c. Upon adding another telephone line, we have

Average arrival rate Ri = 1/4 per minute, Average unit capacity 1/Tp = 1/3 per minute, Number of servers c = 1, Maximum buffer size, K = 4. Using the Performance.xls spreadsheet we get Average waiting time Ti = 4.33 mins, Average # of customers on hold Ii = 1.005, Average number of customers in system I = 1.70, Probability of blocking = 0.072. In this case the costs incurred are the wages of the CSR, the cost of waiting (line charge + waiting cost for customers) and the lost business because of blocked calls. We have Hourly wages of CSR = $20 / hour, Line charge (of existing lines) = $5 / hour, Customer waiting cost = Average number on hold 60 $2 = 1.005 120 = $120.6, Cost of blocking = Calls blocked per hour $100 = Probability of blocking Average arrival rate $100 = 0.072 15 $100 = $108. Excluding the cost of the new line we have Total cost per hour = $20 + $5 + $120.6 + $108 = $253.6. As long as the cost of the new line is less than $273.4 (cost with 4 lines) - $253.6 (cost with 3 lines) = $19.8 / hour, it pays to install the new line.

Upon adding another server (assuming that the fifth line has been added) we have
Average arrival rate Ri = 1/4 per minute, Average unit capacity 1/Tp = 1/3 per minute, Number of servers c = 2, Maximum buffer size K = 3. Using the Performance.xls spreadsheet we get Average waiting time Ti = 0.42 mins, Average # of customers on hold Ii = 0.105 Average number of customers in system I = 0.85, Probability of blocking = 0.007. In this case the costs incurred are the wages of the CSR, the cost of waiting (line charge + waiting cost for customers) and the lost business because of blocked calls. We have Hourly wages of CSRs = $40 / hour, Line charge = $5 / hour, Customer waiting cost = Average number on hold 60 $2 = .105 120 = $12.6, Cost of blocking = Calls blocked per hour $100 = Probability of blocking Average arrival rate $100 = 0.007 15 $100 = $10.5. In this case we have Hourly cost of system = $40 + $5 + $12.6 + $10.5 = $68.1/hour. This is a significant reduction in cost. The new CSR should thus be hired.

Problem 8.5 (Heavenly Mercy Hospital)

a. We have: Average arrival rate Ri = 18 per hour = 0.3 per minute, Average unit capacity 1/Tp = 2 per hour = 1/30 per minute, Number of servers c: To be determined, Cost per server = $100 per hour, Desired average time in system = 40 minutes. To plan staffing, we know that we should have a utilization of less than 100%, thus: Utilization = inflow/capacity = 18/hr/(c*2/hr) < 1 so that c > 9 Increasing the number of servers from 10 upward, we have the following results: (using the Performance.xls spreadsheet with K=100): Number of Servers (c) 10 11 12 Avg. Number in System (I) 15 11 10 Avg. Time in System (T) 50 minutes 36 minutes 32.7 minutes

Thus hiring 11 servers achieves a turnaround time of 36.46 minutes on average. This is under the desired target of 40 minutes. The hourly cost of this system is $1,100.

Now consider the case where the service time is reduced to 20 minutes but the cost of the equipment and radiologist is $150 per hour. In this case:
Average arrival rate Ri = 18 per hour = 0.3 per minute, Average unit capacity 1/Tp = 3 per hour = 1/20 per minute, Number of servers c: To be determined, Cost per server = $150 per hour, Desired average time in system = 40 minutes. To plan staffing, we know that we should have a utilization of less than 100%, thus: Utilization = inflow/capacity = 18/hr/(c*3/hr) < 1 so that c > 6 Increasing the number of servers from 7 upward, we have the following results: (using the spreadsheet (using the spreadsheet with K=100) Number of Servers (c) Avg. Number in System (I) Avg. Time in System (T) 6 Large Large 7 9.7 32.28 The cost of hiring seven servers = 7 150 = $1,050. Thus it is advantageous to lease the more sophisticated equipment. It reduces the cost and reduces the overall time spent in the system.

Problem 8.6 (First Local Bank)

a. We currently have: Average arrival rate Ri = 30 per hour = 0.5 per minute, Average unit capacity 1/Tp = 10 per hour = 1/6 per minute, Number of servers c = 4, Cost of customer waiting = $20/hour = $1/3 per minute. Performance of the current system is (using Performance.xls) Average waiting time Ti = 3.06 mins, Average time in system T = 9.06 minutes, Average # of customers on hold Ii = 1.53. We thus have Hourly cost of customer wait = Average number waiting $20 = 1.53 $20 = $30.6

b. If the new equipment is leased with the same number of servers, we have Average arrival rate Ri = 30 per hour = 0.5 per minute, Average unit capacity 1/Tp = 15 per hour = 1/4 per minute, Number of servers c = 4. Performance of the new system is (using Performance.xls) Average waiting time Ti = 0.35 mins, Average time in system T = 4.35 minutes, Average # of customers on hold Ii = 0.17. We thus have Hourly cost of customer wait = Average number waiting $20 = 0.17 $20 = $3.4, Hourly cost of high-speed equipment = $30 per hour. This implies that Total cost of new system = $33.4 per hour. This is somewhat higher than the total cost of the original system and cannot be justified in financial terms. However performance in terms of customer waiting is significantly improved with total time in system reducing from 9.06 minutes to 4.35 minutes. If the goal is to reduce average time in system below 8 minutes, the high-speed equipment is worth leasing.

Problem 8.7 (Global Airlines)

Average arrival rate, Ri = 52 per hour = 52/60 per minute, Average unit capacity, 1/Tp = 20 per hour = 1/3 per minute, Number of servers, c : To be determined, Cost of customer waiting = $60/hour = $1 per minute, Cost per server = $20 per hour. To plan staffing, we know that we should have a utilization of less than 100%, thus: Utilization = inflow/capacity = 52/hr/(c*20/hr) < 1 so that c > 52/20. Increasing the number of servers from 3 upward, we have (in the spreadsheet, set buffer capacity K =100): Number of Servers c 3 4 5 Server cost per hour $60 $80 $100 Average Queue length Ii 4.95 0.66 0.16 Average waiting time Ti 5.71 0.76 0.19 Waiting cost per hour = Ii 60 $297 $39.6 $9.6 Total cost per hour $357 $119.6 $109.6

Thus Global should staff with 5 agents. Hiring a sixth agent will raise the agent cost to $120 per hour and is not worthwhile. The industry norm of averaging under 3 minutes of waiting can be achieved using only 4 agents.

Problem 8.8 (Henniker Bank)

a. For one district, Ri = 3.5 claims/week = 3.5/5 or 0.7 claims/day, and Tp = 1.2 days, so Rp = 1/1.2 = 0.833 claims/day. With c =1, T = 1/ (Rp Ri ) = 7.5 days. From Littles Law, the average number of claims in the system (for one district) is 0 .7 ( 7 .5) = 5.25 . Out of those 5.25 - 0.84 = 4.41 will be waiting to be processed. The total number of claims across all districts is then 3 4.41 = 13.23 . Hence, utilization = 0.7/0.833 = 0.84. 10 The fraction of claims that take less than 10 days is given by 1 e 7 .5 = 0 .736 . b. If we reduce the standard deviation of the service process by 50% but leave the mean unchanged, 2 CS = 0.25 . We then have (for one district), Ii = 2/(1- )( 1+ 0.25)/2 = 2.76 claims. The 10 total across all districts is then 8.28. The fraction taking less than 10 days is 1 e 5 .14 = 0 .857

Problem 8.9.(Burrito King)

a. Reducing variability will reduce the waiting time b. Server utilization will remain the same, since the average demand and service rates do not change.

Problem 8.10 (V. V. Ranger)

The average waiting time will decrease. Pooling of safety capacity pooling through one waiting line yields lower waiting times because idleness periods are re-distributed.

Problem 8.11 (Master Karr)

Arrival Rate

Service Time

# of Servers

Buffer Capacity

Average Probability of Average Utilization blocking queue length P(block) II

Part

Ri
4 4

Tp

A B a.

1 1

5 6

10 9

79.00% 66.50%

1.25E-02 2.48E-03

1.58 0.51

Waiting cost = ($60/hr)*(average # of customers waiting) = $60 Ii = $94.65/hr Blocking cost = ($50/call)*(average # of busy calls/hr) = $50/call * R * P(block) = $149.93/hr Staffing cost = $15/SR * (# of SRs) = $15 c = $75/hr Total cost = $319.58/hr b. Now the number of servers c increases to 6 while the buffer capacity K decreases to 9. Recalculate the performance with the spreadsheet (calculations above). Waiting cost = $60 Ii = $30.77/hr Blocking cost = $50/call * R * P(block) = $29.78/hr Staffing cost = $15 c = $90/hr Total = $150.54/hr Hence, we should add a server, as it yields a cost savings of $319.58 - $150.54 = $169.04 /hr.

Problem 8.12 (BizTravel.com) a. This is the resource pooling idea in the context of one queue vs. multiple queue. With the CRM software, BizTravel will be able to meet its service guarantee better. b. b.Now, average arrival rate equals the average service rate. This implies long queues and hence BizTravel will pay out $10.00 often during the lunch hour. Problem 8.13 (McBerger)
a. The average waiting time in queue will decrease, because less variability leads to less waiting, by the queue length formula. b. The average waiting time in queue will decrease because shorter service time leads to lower capacity utilization and hence less waiting

9.3 Solutions to the Problem Set

Problem 9.1
a. Given the symmetric shape of normal distribution around its mean, maximum conformance of the output within the given specifications will be achieved by centering the process at the midpoint of the specifications, i.e., at = 32.5 gms. Now if we desire 98% of the output to conform to the specifications, from the Normal tables, the specification limits should be z = 2.33 standard deviations on either side of the mean, i.e., (35 - 30)/ = 2.33, or = 5/2.33 = 1.073 gms. The corresponding process capability ratio is Cp = (35 - 30)/6 = 0.78. b. With n = 12, and = 1.073 as above, we can now determine the ideal control limits on subgroup averages of 12 bottles as: Average control chart: 3 / n = 32.5 (3)(1.073)/= (31.57, 33.34)

Problem 9.2
a. In order to produce 98% of the boxes above 15.5 oz., the process mean must be z = 2.055 standard deviations above 15.5, i.e., = 15.5 + (2.055)(0.5) = 16.53 oz. b. If = 16.53, then the proportion of overweight boxes will be P(X > 16) = P[Z > (16-16.53)/0.5] = P(Z > -1.06) = 0.8554. c. With = 16.53, = 0.5, and n = 9, the control limits on the average weight in a sample of 9 boxes are: + 3 / 9 = (16.03, 17.03). The observed average of 15.9 is below the lower control limit of 16.03, which signals that the mean has shifted below 16.53 due to an assignable cause. They should stop the process and adjust the mean upward. That the average weight is above the minimum individual weight of 15.5 oz is irrelevant. The specifications are on weights of individual boxes, not on average weights.

Problem 9.3
From the 26 observations given, we can calculate the average number of errors per thousand transactions m = 3.3077, which is much better than the industry average = 15. We can then determine the control limits on the number of errors, assuming Poisson distribution, as m 3 m = (0, 8.75). Observe that three observations out of the 26 given exceed the UCL = 8.75. Hence, the process is not in control, even though on average it is better than the BAI standard! The process is not stable, and our estimate of m is not reliable. We first need to stabilize the process by removing assignable causes.

Problem 9.4

a. If the process mean = 515, standard deviation = 5 gms, and sample size n = 25, Control limits are LCL = 3 / 25 = 512 gms and UCL = + 3 / 25 = 518 gms b. Proportion of underweight output is Prob(W < 500) = Prob [Z < (500 515) / 5] = Prob (Z < 3) = 0.0013 or 0.13%. c. If Prob (violation) = 0.13% and the mean is = 503 gms, then must be decreased so that 3 = (500 - 503)/ or = 1.

Problem 9.5
Given: process mean = 6 cm, standard deviation = 0.01 cm, and sample size n = 10, Control limits are LCL = 3 / 10 = 5.9905 cm and UCL = + 3 / 10 = 6.0095 cm Note: customer specifications are irrelevant.

Problem 9.6
If the specifications do not change, improvement in the sigma capability means the standard deviation decreases. In that case, the control band should become narrower.

Problem 9.7
Currently, process mean = 2.2 hours, and standard deviation = 0.8 hours, so probability of processing within 4 hours is P(T < 4) = Prob[Z < (4 2.2)/0.8 ] =Prob(Z < 2.25) = 0.9878 < 0.992, the banks requirement With improved process mean = 2 hours, and standard deviation = 1.2 hours, P(T < 4) = Prob[Z < (4 2)/1.2 ] =Prob(Z < 1.67) = 0.9616, which is worse than before. To meet the banks requirement, the mean should be lowered to 2 hours without raising variability. In that case P(T < 4) = Prob[Z < (4 2)/0.8 ] =Prob(Z < 2.5) = 0.9938 > 0.992, the banks requirement

Problem 9.8
If D is the diameter of basketballs produced, we need Prob(29.3 < D < 29.7) = 0.98. Now we know that Prob (- 2.326 < Z < 2.326) = 0.98, so z = (29.7 29.5) / = 2.326 or we need = 0.086 inches. Hence, Cp = (USL LSL)/6 = (29.7 - 29.3)/6 = 0.7752.

Problem 9.9 a. False. Control limits only assure process stability. They have nothing to do with meeting
customer specifications.

b. True. As sample size increases, standard deviation of sample averages decreases, and the control
bands becomes narrower.

c. True. If control limits are 3 standard deviations from the mean, the probability of sample average
falling outside control limits is 0.27%, regardless of the value of the standard deviation.

d. If the process improves sigma capability, its standard deviation goes down, and the width of the
control band decreases.