AGGREGATE PLANNING 10.1 DECISION OPTIONS 10.2 BASIC STRATEGIES 39-3 AGGREGATE PLANNING COSTS : 10.4 EXAMPLE OF COSTING 10.5 MATHEMATICAL MODELS 10.6 LINEAR DECISION RULES 10.7 SURVEY OF METHODS 10.8 AGGREGATE PLANNING APPLICATION 10.9 @ EVALUATION OF AGGREGATE PLANNING ‘ 10.10 “KEY POINTS i QUESTIONS PROBLEMS : SELECTED BIBLIOGRAPHY : A ggregate planning is concerned with matching supply and demand of outpur over the medium time range, up to approximately 12 months [Link] future. The term “aggregate” implies that che planning is done for a single overall measure of output or, at the most, a few aggregated product categories. The aim of aggregate planning is to set overall output levels in the near to medium future in the face of fluctuating or uncertain demand, As a result of aggregate planning, decisions and policies should be made“con- ceming overtime, hiring, firing, subcontracting, and inventory levels. Aggregate planning determines not only the output levels planned but also the appropriate re~ source input mix to be used. Aggregate planning might seck to influence demand as well as supply. If this is the case, variables such as price, advertising, and product mix might be used. IF changes in demand are considered, then marketing, along with operations, will be intimately involved in aggregate planning. 7 In this chapter, we will be using a broad definition of the term “aggregate plan ning.” There are also narrow definitions which limit aggregate planning to particular types of mathematical formulations. These formulations might include, for example,- 275276 aur si caracery ratatic AND ScHEDULING * Figure 10.1 Hierarchy of ca- pacity decisions, eee of cost minimization. or part supply variables only, a specific obj matical forms. In the broad sense of th lowing characteristics: definition, the aggregate planning problem has the fol 1. A time horizon of about 12 months, with updating of the plan on a periodic basis (perhaps monthly) : 2 An aggregate level of product demand consisting of onc or 2 few categories © product. The demand is assumed to be fluctuating, uncert: or scasonat 3. The possibility of changing both supply and demand variables. A variety of management objectives which might include low inventories, good labor relations, low costs, Acxibility to mercase future ourpat levels, and good. customer service. 5. Facilities that are considered fixed and cannot be expanded. Aggregate planning forms an important link between facilities planning on the one hand and scheduling on the other. Facilities planning determines the physical capacity which cannot be exceeded by aggregate planning. Thus facilities planning extends further into the future than aggregate planning and constrains the aggregate planning decisions. Scheduling, on the other hand, refers to the short range (a few months or less) and is constrained by aggregate planning decisions. While aggregate planning deals with the acquisition of resources, scheduling is concemed with allocsting available Fesources to specific jobs and orders. Thus 2 basic distinction should be made ber tween auquiring resources through aggregate planning and later allocating then through scheduling. ; This hierarchy of capacity decisions is shown in Figure 10.1 and is discussed in more detail in the introduction tw this part of the book. Notice that the deasone proceed from the top down and that there is also a feedback loop from che bettors up, Thus, scheduling decisions often indicate a need for revised aggregate planning and aggregate planning may also uncover facility needs, “Axgregate planning is closely related to other corporate decisions, involving. for flBOX 10.1 example, budgeting, personnel, and marketing. The rela Particularly strong one. Most budgets are based on 2580 Output, personnel levels, inventory levels, pu should thus be the basis for initial budget develop: conditions warrant. Personnel planning is also greatly affected by aggregate planning because the results of aggregate planning include hiring, firing, and overtime decisions. In the service industries, where inventory is not a factor, aggregate planning is practically synonymous with budgeting and personne! planning, Marketing must always be closely related co aggregate planning because the fa ture supply of output, and thus customer service, is being determined. Furthermore, cooperation between marketing and operations is required when both supply and demand variables are being used to determin the best corporste approach to aggre gate planning. Aggregate planuiing should be scen as an activity which is the primary responsibility of the operations function but which requires coordination and coop eration with all parts of the firm, The coordination problem in a typical aggregate planning decision is illustrated in Box 10.1 TYPICAL DECISION PROBLEM Ganaanplodopamecning wanheld acof tumacesiandy Soe catir278 vani sc caPacery Puasa NG AND SCHEOULNG ‘The next section contains a derailed discussion o} demand and supply for aggregate p ment of 2 manufacturing and 5 ns available to modity owed by the develop vat for both -d with exampkss of 10:1 DECISION OPTIONS The aggregate planning problem can be clanfied by a discussion of the various deo 08 options available. These will be divided into two types: (1) those modifying demand and (2) those mo: difying supply. For a more detailed discussion, sce Gal braith (1969), Demand can be modified of influenced in several ways: 1. PRICING, Differential demand in off-peak peri hotel rates, factory disco. rates, Pricing is often used to reduce peak demand or to build up iods. Some examples are matin: rants for carly- of late. and two-for-one prices at restaurants, to level demand through the day, week, IEC MOviC prices, off-season -scason purchases, nighttime telephone The purpose of these pricing schemes is + month, or year 2 ADVERTISING AND PROMOTION. This is another method uscd to stimulate or ini Spe exses smooth our demand. Advertsing is generally timed so 28 to promote demand during slack periods and to shift demand from the peak periods to the Sack times: For example, ski resorts advertise to lengthen thei season, and turkey: Browers adveftise wu stimulate demand outside the Thanksgiving atzd Christm, seasons. ReeNGRLOG OR RESERVATIONS. In some cascs demand is influenced by asking euse ‘oumers co walt for thet orders (backlog) ot by reserving capacity in advance fons Fytow. Generally speaking, this has the effect of shifting demand from peak pert pul © periods with s'ack capacity. However, the waiting time may result alone oe business. This loss can be desicable when the aint is co maximize profi although ! ‘ost operations arc extremly reluctant to turn away customers: backlogs or ree vations are preferred 4. DEVELOPMENT OF COMPLEMENTARY PRODUCTS. ands sometinies try to develop products wl The classic example ofthis approach isa lawn-mower company that begins building | snow-blowers, In the service industry, an example is provided by fastfood texte fants that begin to offer breakfast so as to smooth out demand sad wel more fully. (The service industrics, using all the mechanisms cited above, have gone much farther than theie manufacturing counterpares in influencing demand. That can prob ably be attributed co one emu ' | Firms with highly seasonal do ich have countercyclie seasonal trends. ize capacity cial difference —the inability of service operations to inventory their product.CHAPTER 101 AGGREGATE PLANNING 279 There are also a lar ge number of variables avai ify oe ete eae eae variables available to modify supply through lenis AND FIRING EMPLOYEES. The use of this variable differs a great deal be~ = mpanies and industries. Some companies will do almost anything before lucing the size of the work force through layoffs. Other companies routinely in- crease and decrease their work forces as demand changes. These practices affect not only costs but also labor relations, productivity, and worker morale. As a result, company hiring and firing practices may be restricted by union contracts oF com-, pany policies. One of the purposes of aggregate planning, however, is to examine the effect of these policies on costs or profits. 2. USING OVERTIME AND UNDERTIME. Overtime is sometimes used for short or me~ dium-range labor adjustments in lieu of hiring and firing, especially if the change in Gemund ie considered temporary. Overtime usually costs 150 percent of regular time, with double time on weekends or Sundays. Because of its high cost, managers. are sometimes reluctant to usc overtime. Furthermore, workers are reluctant to work more than 20 percent weekly ‘overtime for a duration of. several weeks. “Un- Yectime” refers to planted underutilization of the work force rather than layofls or perhaps a shortened work week, Undertime can be thought of as the opposite ptt Overtime. Another term for undertime is “idle time.” 3, USING PART-TIME LABOR. In some cases it is possible to hire part-time employees in order to meet demands. This option may be particularly attractive because part- time employees are often paid significantly less in wages and benefits. Unions, of course, frown on the use of part-time employees, because the latter often do not pay ! Gnion dues and may weaken union influence. Part-time employees are, however, ¢S- | Sential to many service operations, such as restaurants, hospitals, supermarkets, and | Geparement stores. These operations are highly dependent on their ability to attract and utilize part-time workers for periods of peak demand. : 44, CARRYING INVENTORY. In manufacturing companies, inventory can be used as @ buffer between supply and demand. Inventories for later use can be built up during periods of slack demand. Inventory thus uncouples supply from demand in mana Bletaring opefations, thereby allowing for smoother operations. Inventory can be dewed as 2 way to store labor for future consumption. This option is. of course, not Wille for service operations and leads to a somewhat different and more difficult aggregate planning problem for them. 5, SUBCONTRACTING. This option, which involves the use of other firms, is sone times an effective way to increase or decrease supply. The subcontractor may supply the entire product or only some of the components. For example, a manufacturer of toys may tlize subcontractors to make plastic parts during certain times of the year, The manufacturer may’ furnish the molds and specify the materials and Methods to be used. Service operations may subcontract for secretarial help, catering : services, or facilities during peak periods.280 rants) camacrry rannieG nance ensennts are very similar to 6. MAKING COOPERATIVE ARRANGEMENTS, TIhcss arrangsinel™ eee ed Rerworks, hospitals Ad services, and hotels y are fully booked subcontracting, in that other sources utilities which ate hooked together through Pp d theit paticnts to other hoxprals for cerzain special nets ainong one another when ower-aharin which 5 of aiclines which shift custo planning problem gall these options, itis clear that the aggregate tw considering all these options itis clear that the apregate planning PONTE extremely broad and affects all parts of the firm The dea) must, therefore, be strategic and reflect all the fir ning is considered narrowly, suboptimization may occur and inappropriate dect ay result, Some of the multiple tradcoffs which should be considered are service level (through back orders or lost demand), inventory levels, stab ‘ labor force, and costs. All these conflicting objectives and tradcofls are sometinies {into a single vost finetion. A mtethod for doing this will be deseribed 1 objectives Waggregate plan- a sions comb the next two sections, 10.2 BASIC STRATEGIES Two pure operations strategies ean be used, along with many combinations in be- ween, to meet Ructuating demand over time, One pure strategy is to level the work force and the other is to chase demand with the work force. With a perfectly level strategy, the rate of re ime output will be constant, Any variations in demand must then be absorbed by using inventories, overtime, part-time workers, subcon= tracting, cooperative arrangements, or any of the demand-influencing options, What hhas essentially beet done with the level strategy is (0 fix the regular work force by sing one of the above 10 variables available for aggregate planning ‘With the pure chase strategy, the work-toree level is changed tw meet de In this case, itis not necessary C0 earry inventory oF to Use any Of the other variables available for aggregate planning; the work force absorbs all dhe chatiges in demand. OF course, these two strategies are extremes; one strategy Makes no cha the work force andl the other varies the work forey directly wie demand changes. fn practice, many combinations are also possible, but the pure strategies help focus on issucs Consider, for example, the cave of a brokerage firm which utilized both strate gies. The data processing department maintained a capacity to process 17,000 trans- is per day, far in excess of the average load of 12,00. This capacity allowed the department to keep a level work force of programmers, systems analysts, computer ‘operators, and keypunchers, even though capacity exeveded demand on many days. Because of the skilled work force, the high capital investinent, and the low marginal cost of additional capacity, it made sense for the data processing departmeut to fol- low this strategy. Meanwhile, in the cashiering department, a chase strategy was being (ollowed. As the transaction level varied, part-time workers, hiring, and layofls were used This department was very labor-intensive, with a high personnel turnover and a low skill level, The manager of the department commented that the high tumover levelCOMPRRISON OF CHASE VERS. was an advantage, since it helped facilitat low demand It ean be scen from this situation that the character y followed, Sasser (1976) tion to the factors shown im Box 10.2. Whi influence che type of priate for low-skilled labor and routine jobs, the level st ate for highly led labor and complex jobs These strategics, however, cannot be prope! entena atv stated. One way to do this 18 t0 re cost, as described in the next section AGGREGATE PLANNING COSTS Most a be more ap given, the followsn AND FIRING COSTS. T!282 vant a4 caracry reas G_AND SCHEDULING Wages plus 50 to 100 percent premium. The cost of undertime is often reflected by : use of employees at less than full productivity. 3. INVENTORY CARRYING COSTS. Inventory carrying costs are associated with aie taining the product in inventory; they include the cost of capital, variable cost of Storage, obsolescence, and deterioration. These costs are often expressed as a pet centage of the dollar valuc of inventory, ranging from 15 to 35 percent per year. This cost can be thought of as an “interest” charge assessed against the dollar value of inventory held in stock. Thus, if the carrving cost is 20 percent and each unit costs $10 to produce, it will cost $2 to carry one unit in inventory for a year. 4. SUBCONTRACTING COSTS. The cost of subcontracting is the price that is paid to 2 subcontractor to produce the units. It is often more expensive to subcontract than to produce units in house, 5. PART-TIME LABOR COSTS. Because of differences in ber Labor will probably be less than that of regular labor. Ald often get no benefits, a maximum percentage of part-time . operational considerations or by union contract. Otherwise, there might be a tone dency to use all part-time labor. However, the regular labor force is essential to the effective utilization of part-time personnel. firs, the cost of part-time though part-time workers labor may be specified by 6. GOST OF STOCKOUT OR BACK ORDER. The cost stockout should reflect the effect of reduced cu : difficult to estimate, but it can be related te the loss of custemer goodwill aid the * possible loss of future sales. Thus we may think of stockout or back-order casts in terns of foregone future profits for taking a back order or for a istomer service. This cost is extremely | Some or all of these costs may be present in any particular ag problem. The applicable costs will be used to “price out"™ the example below, only a few strategies will be pric models, 2 very large n 4 gate planning | altemative strategics. In out. In later mathematical unber of strategies will be considered. 10.4 EXAMPLE OF COSTING The Hefty Beer Company is constructing an aggre; Although several types of beers are brewed at the sizes are bottled, management has decided to use measure of capacity, The demand for beer over the next 12 months is forceast to follow the pattem shown in Figure 10.2, Notice how this demand usually peaks in the summes tronthe and is decidedly lower in the winter. ‘The management of the Hefty brewery would like to consider three aggregate plans: ¢ plan for the next 12 months Hefty plane and several container gallons of beer as the aggregate 1. Level work force. Use inventory to mect peak demands,CHAPTER 10) AGGREGATE PLANNING 283 600 5 3 100 Figure 10.2 Hetty Beer cate beam tee San Feb, Mars Ast eMay June July, Aug. Sept Oct. Now. Ore beer. 2. Level work force plus overtime. Use 20 percent overtime along with inventory as necessary in May, June, and July to mect peak demands. 1 3, Chase strategy. Hire and fire workers each month as necessary to mect demand. To evaluate these strategies, management has collected the following cost and, resource data: if 1. Each worker can produce 2000 gallons of beer per month on regular time. onl | overtime, the same production rate is assumed; but overtime can be used for | | only 3 months during the year. \ 2. Each worker is paid $1500 per percent of regular time. A max of the 3 months. 3, Tecosts $50N) to hie a worker, including screening costs, paperwork, and train= ing costs. It costs $200 to fire a worker cluding all severance and benefit costs. 4, Forinventory valuation purposes, beer costs $1 a gallon to produée. The cost of carrying inventory is assumed to be 2 percent a month (or 2 cents per gallon of beer per month). : 5, Assume that the starting inventory is 300,000 gallons. The desired ending inven- tory, a ycar from now, is also 300,000 gallons. All forecast demand must be met; no stockouts are allowed. jonth on regular time. Overtime is paid at 150 mim of 20 percent overtime can be used in any The next task is to evaluate cach of the three strategies in terms of the costs which have been given, The first step in this process is to construct charts like those. shown in Tables 10.1 through 10.3, which show all the relevant costs: regular work force, hiring/firing, overtime, and inventory. Notice that subcontracting, part-time labor, and back ordering/stockouts have not been allowed as variables in this case.ea TAMU 101 RT DT CAPACITY PLA! AGGREGATE PLANKING COSTS. Level work force Jan. Resources ? 171 Regutar workers: Tay in im wn Overtime (percent) — - - - Units : produced 342 M2 342 uz ue Sales forecast 200 250 300 400 450 Inventory (end of month) 442 534 578 sia 410 Costs Regular time $256.50 $256.50 $256.50 $256.50 $256.50 Overtime = 7 = = i Hirertire = S ee oa Inventory 8.84 10.68 11.82 10.36 8.20 Total cost $265.34 $267.18 $268.02 $266.86 $204.70 ee eee emo a * All costs are expr ln thousands lm thousands of dollars. All production, Inventory, and asles figures are allons. Starting Inventory Is 300,000 gallons in each cave. TADLE 10.2 AGGREGATE PLANNING COSTS—STRATEGY 2* Use of overtime Jan. Feb. Mar, Apr. May Resources Rogular workers 163 163 163 163 163 Overtime (percent) 20 Units : produced 326 326 326 326 301 Sales forecast 200 250 300 400 450 ‘ Inventory (end of month) 426 502 528 454 395 Costs « Regular time $244.50 $244.50 $284.50 $244.50 5244.50 Overtime 73.35 Miredfire = - - = a Inventory B52 10.04 10.56 Total cost $253.02 $254.54 Prana st trussed in thourands of dotars. Al production, inventory, and sates figures ave 'm thousands of gafuis. Starting Inventory is 300,000 gallons in each eave,CHAPTER to! AGGREGATE PLANNING 285, TADLE 10.1 (Continued) june = end Aug. Sept. oct, Now. Dec. Total im } 7. Ww mn wm 71 m 34 2 342 342 342 342 342 342 4104 500 500 400 350 300 250 200 4100 252 94 36 2a 70 162 308 $256.50 $256.50 $256.50 $256.50 $256.50 $256.50 $256.50 53078.00 5.04 1.88 Ez 56 1.40 3.24 6.08 68.52 $261.54 $258.38 $257.22 $257.06 $257.90 $259.74 926258 SIae EE TABLE 10.2 (Continued) t June July ‘Aug. Sept. Oct. Nov. Dec. Total 1 | 163 163 163 163 163 163 163, 20 20 | 391 391 326 326 326 326 326 4107 500 500 400 350 300 250 200 4100 286 7 103 79 105 181 307 $244.50 $244.50 $244.50 $244.50 $244.50 $244.50 $244.50 $2934.00 73.35 (73.35 220.05 5.70 3.54 2.05 1.58 2.10 3.62 6.14 70.84 $323.55 $321.39 $246.56 $246.08 $246.60 $248.12 $250.64 $3224.89an are eR > 2BG rant) capac TABLE 10.3 ee SCHEDULES AGGREGATE PLANNING COSTS_steAtESY Crate demane den. Feb, Mar. Aer. = Revaucees Regular workers 190 Rs 180 20 a Overtime percent) = = - - Unite reduced 200.28 300 a8 Sates Vorecatt 20 50a er) Inventory {end of month) 300300 300 30 (300 Costs Regular ime $150.00 $187.50 $2500 30000 337.50 Overtime = = = = = Nirere = 1250128050080 toventory 09 SO Totalcon $156.00 $706.00 F243.50 01.00 SEO Inventory, ond ates figures are ‘ach cave {n evaluating the first option, we mast calculate the size of the work force res quired to meet the demand and inventory goals. Since ciding and beginning iav tories are assumed to be cqual, the work force must be just large enough to mect total demand during the year. When the monthly demands from Figuec 10.2 are added, the annual demand is 4,100,000 2000(12) © 24,X7 gallons in a year, a le workers is needed co mect the total demand. On the basis of this we the inventories for cach month and the resulting Table 10.1. ‘The second strategy is a bit more complicated, since some overtime can be used. IN is the work-force size for option 2, we must havs 92000) + 3{(1.2(2000X)] = 4,100,000 For 9 months we will produce at 20X0X gallons per month and for 3 months we will produce at 120 percent of this rate including overtinie. Whex the above equation is solved for N, we have N = 163 workers on regular tinne. In Table 10.2 again calculated the inventories and resulting costs for this option, The third option requires that the work force be varied in each month to a the demand by hiring and firing workers. Straightforward calculations produce the number of workers and associated costs for cach month. In notice that a constant level of 300,000 units in inventory is maintained 2s the mininvain ineentory level The annual costs of coch strategy are collscted and summarized in Box 10.3 gallons. Since cach worker can produce cl work force of 4,100,00024,000 = 171 | ork-force figure, ts have been calculated in we have oncerab 103 (Continued) accnecate pranninc 287 june a ame ay Aug. Sept. Oct. Nor. Dec. Tetal 250 250 200 75 150 125 100 500 500 400 350 300 250 200 4100 oo S00 400 350 300 250 200 4100 300 300 300 300 300 300 300 $375.00 $375.00 $300.00 $262.50 $225.00 $187.50 $150.00 $3075.00 ue 12.50 - 10.00 5.00 5.00 5.00 5.00 105.00 : 6.00 6.00 6.00 6.00 6.00 6.00 __ 6.00 72,00 |) 335350 HseL.00 © Fiso0 © sz7a.50 $236.00 Sise.s0 i600 $5252.00 Based on the assumptions used, strategy | is the least-cost strategy, but cost is not the only consideration, For example, altemative 1 has more flexibility to respond to higher demands since no overtime has been planned. Altemative 2, on the other, hand, has less down-side risk, since planned overtime can be cancelled. If prospects, for higher-than-foreeast demand are good, alternative | may be preferred; if lower than-forecast demand is more likely, alkemative 2 may be preferred. BOX 10.3 cost SUMMARY PPR Prkr erty ECEER Li FERAL} pO EEA) iy288 rane si caracere rarine 10.3 Cumulative de- ‘mand and produc- tion, D SCHEDULING Jin, Fab, Mar, Apr. May dune July Aug Sept. Oct. Now. Dee Monts 1 useful to compare cumulative demand against cumulative production plus initial inventory, 28 shown in Figure 10.3. The; difference between these curves will then be the inventory (or backlog) at any given | time. The inventory buildup rates and inventory levels can be scen at a glance fom | these cumulative curves. We have shown how to co pare costs in a very simple case of aggregate plan- ting for a few strategics. In Section 10.5, many other strategies and more complex cost structures will be considered by formulating the aggregate planning problem as 4 seties of mathematical models. 10.5 MATHEMATICAL MODELS A side varity of aggregate planning models have been developed over the years. In this section the general model formulation will be described. This formulavon will be followed by a discussion of incar decision rules and a survey of the entre range of modeling approaches available. gi? formalize the aggregate planing problem, the following symbols will be use P, = production level during period f, output in units W, = work-force level during period 1, in mumber of s-orkerscoarten 19 aganeaate ruareanc 289 Ty = inventory at the end of penod ¢, in units { = demand forecast for period, in units Assume that demand forecasts are given for cach period, Fi, Fa, «> » Fr. over the future planning horizon T. Abo assume the inital production level Py, work- force level H,, and inventory level I, are given. The problen isco find the values of P, and W, which will meet management’s objectives, usually assumed to be mint mum cost, while meeting the given demand Inv using an aggregate planning model, only the values of Py and Wy for the first period are implemented. Although the succeeding values of Px. Ps. » Py and Wi, Ws, . Wy are specitied by the model, these subsequent valucs will not be implemented unless conditions remain exactly the same for the next period The intodel is thus used sequentially, with decisions from only one period at 2 time being implemented. In addition to providing the best decisions, aggregate planning models often provide decision rules which can be used to calculate optimal decisions in successive Rime periods. Typically two decision rules are provided, one for Py and another for We The cost parameccts, intial values, and demand forceasts serve as input for these degision rules and are used to calculate P, and IV; in any time period. We will ilus- trate several decision rules below by using P, as an example. ‘One decision rule for P, would be to simply let =F In this case the production level would fluctuate with forecast demand, This is the \ chase rule which has been described previously. ‘Another decision rule for P; would be to let P= Pra + AE ~P where A is a smoothing constant, (SAS [Link] decision rule would smooth the \ value of P, depending on the choice of A ina manner exactly analogous to exponen- Talemoothing, IfA = 0, we will have a level strategy, with no response to demand, Gnd if. A = 1, we will have the chase strategy. Thus a varicty of smoothed responses can be obtained siniply by choosing a value for A. Pothaps the value of P, should also be affected by inventory levels. In this case let Py = Pry + AUFe = Prot) + Bly = bea) where [, = normal inventory level and B = smoothing constant, 0 = B S 1. Here, Praill be decreased if inventories are above the normal level; otherwisé Py will be wncreased. Again, the rate of response ean be controlled by choosing B. Finally, the decision rule can be enhanced by considering not only Fy but all fu- ture demand forecasts, Fie1. Fras. cte. At this point, the decision rule will appro: mate rules derived from actual aggregate planning models. By adjusting the coeffi- Tames in these rules, one can achieve any desired response between a chase and 2 level strategy, A cost structure is necded, however, in onder to choose the “best” rule; this will be described next,290 any st caracty uatine ann soHtnyune 10,6 LINEAR DECISION RULES The angicpate planning problem was first formulated by Holt, 4 Simon in 1955. Their formulation utilized quadratic costs and resulted in the so- called linear decision rule (LDR), which was applied to a paint factory problem. The LDR formulation assumed four types of costs Modigliani, and 1. REGULAR PRODUCTION COSTS. The cost of regular-time’ production in period was assumed to be Coc Notice that the cost of regular production was linearly related to the size of the work force, as shown in Figuire 10.4a, 2, HIRING AND FIRING COSTS. The cost of increasing or decreasing the work force in petiod ¢ was assumed to be Q CH) = Cx — Wee Figure 10.4 Costs for linear decision rule, (9) Regular produc- tion cost. (6) Hir- Ing and firing cost. {c) Cost of over- time. (d) Cost of Inventoriesback orders. Work fores (man-montha) wo Net Inventory (units of product) te . w Produetlon rete (unts/month) — — eusaticcont nodel : etn! cont trom company records| Ccharren ms accnecaTe rrannine 291 | was a squared function of the amount of in~ This function was an approximation to the costs observed in the paint factory, as shown in Figure 10.46. The quadratic form was chosen for mathematical convenience, as an approximation. The cost of changing the work fore crease or decrease in the work forc 3. OVERTIME COSTS, The overtime cost was expressed as zero cost up to 100 percent utilization of the work force and then as a linear cost for overtime production be~ yond 100 percent. See Figure 10.4¢, Through the use of quadratic functions, this Overtime cost was approximated as follows: Calt) = Ca(Pr — Ca? + CsPe — Cols 4, COST OF INVENTORIES AND BACK ORDERS. In the LDR formulation, back orders were treated as negative inventory. The following inventory/back-order cost func- tion for the paint factory was used, also see Figure 104d. Cal) = Cle — Ca — CoP The total cost in each period was just the sum of the above four costs: CH) = Ci) + Cx) + Cale) + CD and the cost over the planning horizon ‘T was the sum of the period costs: \ c=3cw \ The LDR problem is to minimize the toral cost C by choosing IY; and P; for cach’ period. These variables are connected from period to period by the inventory bal=} ance. ; he hat PF i! When I; is substituted in the above cost equations, the resulting problem is an uncon- strained minimization problem which can be solved by using calculus, Since all cost functions are quadratic, the optimal solution to this problem is a lincar decision rule for P; and W. In general, these rules may be expressed in the following for Py = AdFr + AsFies + + + + ArFrer + biWea — dilate W.= AsFy + AiFart 00+ + AbFur + bi% where Ay, Als bi. bis dad), and ey, ¢f are constants. As may be noted, the two deci- sion rules have exactly the same form, only the coefficients vary in magnitude. Each of these rules is linear in the variables F, IV, and rule." In the paint factory, several interesting results were observed. The production level was greatly affected by the sales forecast for the next 3 months; the remaining 9 months had lite effect. The work-force level was much less affected by short-term sales forecasts; instead, it depended on the entire 12-month forecast, Thus short= ence the name “linear decision292 rant > CAPACITY PLANNING AND ScHENUL iG ji vhile tange changes in demand were absorbed by changing the production rate, w longerange change were absorbed by work-force changes. These rules, therefore, ; had a great deal of intuitive appeal. : “ The advantage of the LDR formulation és its simplicity and case of use. There ate, however, several disadvantages 1. The LDR is restricted to the use of quadratic cost approximations. More general cost forms cannot be used. The LDR reacts gradually to changed forecasts, hiring or firing a few people cach month. Management somictimes prefers to hold off changes in the work force and then, when changes are required, to make them in larger increments. 3. Since no limitations are placed on the variables, a negative work force could be employed, or inventory could be carried beyond warehouse capacity. This re~ sult, however, is remote when the dezision rules arc used in their normal range. Despite these limitations the LDR has considerable historical value and continucs to be used as a benchmark for comparison with other methods. 10.7 SURVEY OF METHODS Since 1955, the aggregate planning problem has been a rich source of research aimed 3 removing restrictions in the LDR formulation. To provide a historical perspec= ve, eight of the methods will be sumniarized in the order of their development; The purpose of these summaries is not to concentrate o:1 the details of any particulse method but to provide an appreciation for the wide varicty of possible approaches, More complete surveys arc given by Silver (1967) and Eilon (1975). a 1, TRANSPORTATION METHOD, In 1956 Bowman suggested that-the transportation { | method be used to solve a special case of the apgregat: planning problem, In his formulation, the rows represented the available supply of regular time, overtime, and subcontracting in cach month. The columns represented the demand required, ‘ Production could be scheduiled in a current month to satisfy demand in a later wonth by carrying the production in inventory. Although the problem was easy to selve through the transportation method, it did no" include hiring oF firing cost, or back. costs could result in indiscriminate hiring and firing, ch is unrealistic in practice. 2 LINEAR PROGRAMMING. In 1960) Haussmann and Hess formulated the aggregate Planning problem in a linear programming format. The problem is similar in cecry aspect to the LDR except that lincar costs arc used in place of quadratic costs, This formulation also allowed the use of constraints, such as a maximum amount efover, time and inventory and it allowed the use of the powerful simplex method for solus tion and sensitivity analysis. The major restriction is the requirement that linesr'er Piscewise linear costs be used. Nevertheless, this approach relieves the restrictions of +CHAPTER 10 | AGGREGATE PLANNING 293 | the transportation method, and it will yield better results than the LDR when the costs arc approximately-linear. See the Supplement to this chapter for details. 3. MANAGEMENT COEFFICIENTS. In 1963 E. H. Bowman suggested that past manage- ment behavior be used to determine the appropriate coefficients of the production- level and work-force decision rules. This was quite a departure from previous think- ing, which utilized costs to sct the value of these coefficients. Bowman's logic was based on the assumption that management makes good decisions but could perhaps make them more consistent by using a mathematical decision rule. His rule has been criticized by Eilon (1975) and others for being second-order in nature and self-de- feating, since a good tule is obtained only when management decisions have been consistent, and this is exactly the case where the rule is not needed. Because of these difficulties, all other formulations have continued to use cost, profit, or multiple ob- Jectives—rather than past management performance. 4, SIMULATION. Vergin (1966) supplied an example of how simulation can be used to select parameters for aggregate planning decision rules. Through the use of simu- lation, any desired cost structure or other objectives can be evaluated. A disadvan- tage is that only a limited number of rules can be evaluated, since each rule requires a separate simulation run, A heuristic search method can be used, however, to provide a systematic basis for selecting alternate runs. Historically the use of simulation repr resents a step away fron: the highly restrictive linear ané quadratic cost formulation§ of previous methods. The simulation method should be considered when complex, cost structures are required. 5. PARAMETRIC PRODUCTION PLANNING. This method was developed by Jones in} 1967, also to allow more general cost functions to be used. In it, the two LDR rules | ; are generalized by the introduction of two smoothing parameters in cach rule. The micthod utilizes a grid search over the four-dimensional parameter space to choose the parameter values which minimize costs. The resulting rules work quite well when tested against the LDR or linear programming methods. 6. SEARCH DECISION RULE (SOR). In 1968 Taubert introduced his SIR method to find the best rule for generalized cost structures. This method represents a refine- ment of previous formulations in that any desired cost funetion and decision rules can be used. The resulting cost function is minimized by using the Hookes-Jeeves pattem-scarch method. This method establishes a search procedure based on past patterns of step size and direction which have reduced the cost function. Through successive steps, the procedure can arrive at an approximate local minimum of the cost function. Taubert used the procedure to evaluate decision rules with 20 vari- ables and obtained very good results in a reasonable amount of computer time (less than 20 seconds), 7. GOAL PROGRAMMING. 1d Moore (1974) suggested a poal-programming for-294 rant 31 CAPACITY PLANNING AND SCHEDULING sch as the fol- mulation for aggregate planning. In this formulation, multiple goals such as the lowing are specified in priority order: P, = operate within the limits of productive capacity P, = mect the contracted delivery schedule Py = operate at 2 minimum level of 80 percent of regular-time capacity P, = keep inventory to 2 maximum of three units P, = minimize total production and inventory costs Py = hold overtime production to a minimum The solution procedure secks satisfaction of these goals, starting with P; and pro- ceeding to Pz, Ps, etc. Through the use of this approach, tradeoffs can be made between the goals of capacity, delivery schedule, stable work force, production, in~ ventory, and overtime cost. This makes it possible to consider a framework that is broader than the simple cost structurcs of the past. 8, PRODUCTION-SWITCHING HEURISTIC. About the time that some authors thought that all methods had been exhausted, Mellichamp and Love (1978) developed the production-switching heuristic. Their method is based on the observation that man- agers scem to favor onc large change in work force over a scrics of smaller and more frequent changes. Accordingly, 2 three-level—low, medium, and high— produc tion (and work-forcc) rule was formulated. Production is switched from one level to another depending on the level of inventory and sales forecasts. The production- switching points are based on minimization of any given cost function through the use of a scarch procedure. The advantage of this method is that Auetuati petiod-to-pcriod production are held to a minimim, ions in 10.8 AGGREGATE PLANNING APPLICATION | uy To illustrate the use of the above methods, an appl be review cation of aggregate planning will t, The application was done by Taubert (1968) in a research laboratory which employed approximately threv hundred scientists and a hundred support ew ployees. Rescarch for both company projects and outside government done. However, there was considerable fluctuation in the load on the to the changing nature of government contracts The aggregate planning problem was to determi tific and support staffs as well as an allocation of th ment contracts, company research programs, and overhead. The purpose of the overhead allocation was to retain scientists temporarily until future work developed The aggregate planning problem was therefore formulated in terms of the following four variables for cach month ofa G-month planning horiéon: contracts was laboratory due cach mouth the size of the sci ¢ scientific staff to govern- Staff allocated co government contracts Staff allocated to corporate research programs Staff allocated to overhead ; Size of support staff sereFigure 10.5 conrrin wp accarcare reaston 295 Taubere [Link] cost structure of this Laboratory in terms of the 12 cost functions, shown in Figure 10.5. These cost functions included linear, piecewise lin tat, and nonlinear costs, The objective was to choose the 24 variables which would fuiaimize the overall cost. The minimization was achieved by using the scarch meth- odology described above. “Teubert validated the model over a S-year period. When compared to manage- iment decisions, the model's results showed fewer Auctuations in the work-force level from month to month. More importantly, the analysis indicated that costs could have been reduced by 12 percent through use of the model It should be noted that Taubert’s search mode! is very similar to the manufac turing problem execpt tha gorics. leis inte hhas no inventory and it uses multiple manpower cate esting that four types of staf were used instead of one overall aggre~ gate manpower category. This example illustrates the need (0 Gilor a model to the specific decision problems faced by management296 rant ss carncery pias AND SCHED! NG 10.9 EVALUATION OF AGGREGATE PLANNING TABLE 10.4 Now that 2 large number of methods have been reviewed, what can be said rel these methods and their use in business? First, it should be observed that most of the methods incorporate the concept of a decision rule. This concept is fundamental to aggregate planning and should be clearly understood. Second, there has been a tendency over time for the methods to become more powerful in handling general cost structures. While the carly methods were re stricted to quadratic and linear costs, more recent methods can handle any desired cost fianction, The more recent methods have also been designed to handle multiple criteria through goal programming of simulation methods Finally, there have been changes in decision rules. While the early rules were strictly of a smoothing nature, more gencral decision rules are now possible. Exam- ples of the more general rules are the parametric production-planning formulation and the production-switching heuristic. Although’ has been 2 vigorous development of methods, there have been only a few studies ofhow well these methods perform in practice. One study of note is the original LDR paint factory study, which showed a possible cost reduction of 85 percent. Twenty years later, however, Schwarz and Johnson (1978) showed that ‘tuuch of this'reduction was attributable to aggregate inventory policy, not detailed aggregate planning rules. i In three companies, Bowman (1963) showed that the LDR could have reduced costs by 1.5 to 3.3 percent over management performance. Lee and Khumawala (1974) report on a study comparing several decision rules with management practice ina job shop. The results, shown in Table 10.4, indicate that profits could have been improved 13.6 percent by the best decision rule. over $600,000 per year. Although there is good evidence that aggregate planning methodologies can imts | prove profits or reduce costs in practice, the insplementation record has been poor: | ‘An ongoing usc of aggregate planning methods is rarcly reported in the literature Even in the original paint company, only 2 part of the LDR was implemented, In most other cases, the company was used as an example of what could have been done, not what actually was done over a period of time. Why has this occurred with 4 methodology which is rather well developed and apparently has the potential to This was an impressive saving of COMPARATIVE PROFIT PERFORMANCE Imperfect forecast Company decisions $4,420,000 near decision rules 4,821,000 ‘Management coefficients model 4,607,000 Parametric production planning 4,300,000 Search decision rule 5,021,000 ‘Source: W. 8. Lee and B. M. Khumawala, "Simulation Testing of Aggregate Production Planning Models m | an implementation Methodology.” Management Science, val 20, Fesrusry 1374, 908,—— ze manne 297 improve profits substantially? Shouldn't man: methods? be anxious to adopt these nethods? A study done by Shearon (1974) helps to shed some lig ese questions (Also sce Colley, Landel, and Fair (1977), chapters 7 and 8 J in his survey of aggre gate planning in industry, Sharon surveyed 100 companics from wht 48 replies. These companies were all included in the Fortune 500 hist industrial companies in the United States The results of Shearon’s survey indicate that production managers are typically responsible for aggregate planning decisions but general managers often review and approve large changes in inventories or work force. The study also indicates that the aggregate decisions arc fragmented, with marketing controlling variables which in- fluence demand and operations controlling supply variables. There is little coordina- tion between marketing and operations strategics Of the 48 firms surveyed. 42 indicated that they preferred to maintain a level work force whether demand was fluctuating, seasonal, or uncertain. This preference scemed to be influenced by continuing efforts from labor unions to make labor 3 fixed cost. In half the cases, the unionized firms were required to pay supplemental unemployment benefits. These benefits ranged up to 90 percent of take-homte pay for up to 26 weeks while a worker was on layoff, making layoff and rehire very expensive propositions. ‘Operations managers were also asked how they were evaluated on job perform= ance. Meeting schedules Was clearly the most important criterion, followed in order, of importance by controlling direct costs, controlling indirect costs, inventory tum- ‘over, and labor relations, These criteria seemed to help explain the types of aggre~" gate decisions which were made, In periods of increasing demand, operations man- ayers added overtime first, followed by increases in the work force. The typical | manager will use up to 8 hours overtime per week for 12 weeks before adding em: ployees. Reductions in inventories or increased back orders are clearly secondary op~ | tions when demand increases. On the other hand, when demand decreases, inven- tories are allowed to increase and layoffs in labor are considered only asa last resort. This behavior can perhaps be explained by the greater emphasis on schedule than on costs oF inventory tumover. Ic is interesting to note that management behavior on the up side of demand is different chan on the down side. It is also interesting ¢o observe the fragmented na- ture of aggregate decisions with multiple criteria rather than a single criterion such as cost. Vollmann (1973) also sheds some light on aggregate planning in industry by his study of a furniture manufacturer. In this case Vollmann was part of a research team. which developed an aggregate planning model similar to the Search model Al- though management was very pleased with the results, they were not inclined to use the model. It appeared to management that other short-range scheduling problems were more pressing. To gain user support, the research team then shelved the aggre- gee planing model and proceed to develop a computerized scheduling system. ing systemt was successfully implemented. Upon returning to the aggregate planning problem, Vollmann and his associates he received the largestthe managers still did not want to use the original model oe eat ly do aggregate planning in the classical sense. First cies eo “e product groups, since these groups were produced ng was not done by one overall measure of aggrenate nt did not think thae all their objectives could be com le cost ebjective: they fei, instead, that the objectives should be fy. As a result, the rescarch group developed a simulation model problem faced by management. | projected inventories and back orders into the future for 36 months 6. using demand forecasts and current work-force, overtime, and ng levels. The model permitted management to evaluate suggested anges im capacity variables in terms of their effect on projected inventory and é is, and the model was programmed for computer time sharing so easily manipulate the variables and receive the results. Manage- ment may not percvive the existence of an aggregate planning problem, ‘work force and inventories may be made on a reactive week! rough scheduling. If this is the casc, management will nced to lanning function and assign the respons before quantitative techniques will be useful : ot utiderstand the value of a quantitative approach, In Voll, | < Lack of familiarity with quantitative approaches required! | before the more general aggregate planning, ni argues that modeling efforts must be oriented | 4 user problems if implementation is to succeed. egite plzaning models should be tailor-made for the particular situation. It sary to ticlude more than one aggregate product type in the model, product allocation decisions between plants, or to work with not only costs. The management problem should always ly first and aleruate formulations should be explored, ¢3 particular model co the situation a some companies, segregate planning is highly constained by a policy such as cnance of 3 level work force. In sonte cases, the aggregate planning prob- be viewed as 2 one-time policy-cvaluation effort rather than an ongo- ing mode! for making monthly decisions. Before a company accepts an 2ggregate planning approach, the modei's capacity to pro ‘t decisions should be demonstrated. This can often be donc by comparing past management performance with the results which could have thieved by the model. If the results are not improved, a better model dé be built of the present methods continued in use. shaL. care anno 299 Caastes Ik should be possible to umprove agerogste pT nict, The potential improvement has been show = dard techniques have been strate that practice can. is 10.10 KEY POINTS QUESTIONS Aggregate planning serves 25 a link between and scheduling. The aggregate planning decision ne range. AS a result, decisions regarding ageres: subcon- tracting, and back-order levels are also level of facilities available, and they constra for scheduling ‘The following key points are made in the chapter + Aggregate planning is concemed with matching supply and demand over the medium time range. In the aggregate planning problem, the overall output level is planned so as to use the best possible mix of resource imputs + Supply variables which may be changed by aggregate planming are hiring, Firm, overtime, idle time, inventory, subcontracting, part-time labor, and cooperative Sreangements Variables which are available to influence demand are pricing, promotion, backlog or reservations, and complementary products © When demand is given, two pure strategies are available for adjusting supply: the chase strategy and the level strategy. There are also many mixed serategics be~ tween these two extremes. A choice of strategy can be made by determining the total cost of cach of the strategics available. Many models for aggregate planning have been proposed. The first model to ba developed was the linear decision rule, which munimized 2 quadratic cost func~ tion by using two decision rules: one for work force and the other for production level. Since d the new models have generalized the cust structures which can be used and have included various other types of decision rules «Despite the number of models available and the favorable results in a few cases, aggregate planning models have not gained widespresd acceptance ia industry. A more concerted implementation effort may be nceded which includes careful def inition of the decision problem in cach case, tailor-made models, and demonstra~ tion of improved planning results. 1. A definition of aggregate planning by another author is given here. Contrast and compare this definition ta. the one given in the text. “Aggregate planning or scheduling refers to planning in general terms—considering product groups, special promotions, trends in workforce availability, changes in resource suppliers, etc. The objective of aggregate planning is to effectively allocate system capacity —men, materials, and equipment—over300 Pane 51 caracrry ruate PROBLEMS A 12 2B. 1 2. ING AND SCHEDULING tome time horizon.” (Richard B. Chase and N. J. Aquitenn, Production and Operations Management, rev. ed.. H Aggregate planning is some . The XYZ Company manufactures a seesonal product, At the present time, they use 9 fevel labor force as a matter of company palicy. The company is afraid that, If they lay off work ers they wl not be able torenve them cr to ind qualified replacements, Does this cont. pany have an aggregate planning problem? Discuss. : Tehaz been said that aggregate planning Is reated to personnel planning, budgeting, and market planning. Describe the nature ofthe relationship between these types of planning. Every company has multiple objectives such as good laber relations, low operations costs, high Inventory turnover, and good customer service. What are the pros and eons of treating these objectives separately In an aggregate planning problem versus combining them all Into a single measure of cost? How is the choice between a level strategy and a chase strategy influenced by the shill level of the work force and the degree of automation? After all, isn’t the choice between these strategies just a matter of the lowest cost? What factors are important in choosing the length of the planning horizon for aggregate planning? : What are the cost factors that should be Included in calculating the total cast of an aggre gate strategy? . Provide a definition of a decision rule. Is a decision rule always required In an aggregate | planning model? . What assumptions are used in deriving the LDR model? - What are the advantages and disadvantages of the following aggregate planning methods? 2a, Linear decision rule (LOR) f. Parametric production planning ». Transpertation method &. Search decision rule (SDR) ¢. Linear programming bh, Goal programming d, Management coefficients 1 Production-switching heuristic e. Simulation : What problems are being experienced in using aggregate planning in industry? How can these problems be overcome? A barbershop has been using a level work force of barbers 5 days a week, Tuesday through Saturday, The barbers have considerable ile time on Tuesday through Friday, with certain peak periods during the noon hours and after 4 p.m. each day. On Friday afternoon and all day Saturday, all the barbers are very busy, with customers waiting a substantial amount of time and some customers being turned away. What options should this barbershop con- sider tor aggregate planning? How would you analyze these options? What data would be collected and how would the options be compared? 1 ‘The Zoro Company manufactures fast-cutting lawn mowers. The company has devised the following decision rules baged on the LDR methodology: SF, + Wir — 4h BW 4 + OF ~ ey +3 P, w% If, at the present time, Zoro has 100 workers and an inventory of 1000 units, what will be the production and work-force levels for the next 3 months with forecast sales of 800, 1000, and 1200 units? ‘The Chewy Candy Company would like to determine an aggregate production plan for thi next 6 months. The Company makes many different types of candy but feels it can plan itscars nanvos 301 CHAPTER toee! roduc ia Pounds provided the mix of candy sold does net change toe erasticaly Inventory. Each 1, the Chewy Company has $0 workers and 10,000 pounds of cane 1 cues tee worker can produce 100 pounds of candy » month and is pri 45 an hour rote is regular time per month). Overtime, at 4 pay rate of 150 percent of remu! moan ¢ used up to a maximum of 20 percent in addition to regula time s any even It costs 50 cents to store a pound of candy for a yeat, $200 to frre a wanker and $500 to fire a worker. The forecast sales for the next six months are 8000, 10,009, 12,000, 8000, 6000, and 5000 pounds of candy. 1a. Determine the costs of a level production strategy for the next 6 months, with an ending Inventory of 10,000 pounds. 'b. Determine the costs of a chase strategy for the next 6 months. c. Calculate the cost of using the maximum overtime for the 2 months of highest demand. “d. Formulate this as a linear programming problem. 3, Draw a cumulative graph of demand and of the three production strategies given in Problem 2. 4. A company produces to a seasonal demand, with the forecast for the next 12 months 33 given below. The present labor force can produce 500 units per month. Each employee added can produce an additional 20 units per month and is paid $1000 per month. The cost aaa stectate fz $20 per unit. Overtime can be used at the usual premium of time and # half for labor up to a maximum of 10 percent per ‘month, Inventory carrying cost is $50 per unit (err geat. Changes in production level cost $2000 per additional 20 units due to Nine, fring, line changeover costs, etc. Assume 100 ‘units of initial inventory. Extra capacity may be aotaned Ey subcontracting at an additional cost of $15 per unit over and above producing coeatresurteit on regular time. What plan do you recommend? What is the incremental cost of this plan? Pe Ao Demand 600 700 800 700 600 500 600 700 800 900 700 600 5. Approximately 40 percent of a medical clinic's weekly incoming calls for doctor's appoint: tere occur on Monday. Due to this large workload, 30 percent of the callers receive a busy Signal and have to call back ater. The clinic now has one clerk for each two docters to han ser ncoming calls. Each clerk handles calls for the same doctors all week long and thus is te ise with the doctors’ hours, scheduling practices, and Idiosyncrasies. Consider the fol- Towing alternatives to solve this problem: tome tinue the present system, which results In some customer inconvenience, loss of Cosiness, and perceived poor service. About 1000 patients attempt to call the clinic on Mondays. The clinic has 50,000 patients in all. «Expand the phone lines and add more people tohandlle the peak load. Estimated cost to dd five more lines and five clerks is $50,000 per year. 4 Install a computer to speed up appointments. In this case, the peak load could be handled with present personnel. Estimated cost to lease and maintain the equipm grams is $25,000 per year. becetatatar + Expand the phone lines and ask people to call back later in the week for an appoi [Add five lines and five phone-answering clerks part time at $30,000 per pscaitinaed 2. Analyze these options from the standpoint of an aggregate pl tiem, Wha are the pros and cons of each option? ek : 7 b. Which option do you recommend? Why? ‘c. How does this problem differ fom the oth i other aggregate planning problems gi 2 6. Te Revel tel n Oa, Florin, woul ef prepare an aggregate pan fe the nest year, The motel has a maximum of 200 rooms which are fully utilized in the winter months + Advanced problem skills are required.302 PART 3 CAPACITY PLANNING AND SCHEDUL! bbut largely vacant in the summer, 85 shown by the forecast below. The motel requires one employee, paid $600 per month, for each 20 rooms rented on regular time. It can utilize UP to 20 percent overtime at time and a half and can also hire part-time workers at $700 per month. The regular-time workers are hired at a cost of $500 and fired at a cost of $200 per worker. There is no hiring and firing cost for the part-time workers. Month J oF M A M 3 ty A S O N BD Demand 185 190 170 160 120 100 100 80 100 120 140 160 (rooms) a, With a regular work force of six employees and 20 percent overtime when needed, how | many part-time workers ate required in each month and how much does this strategy cost per year? b. What is the best strategy to follow if a level work force of six regular workers is used? You may use various amounts of overtime and part-time workers. *c. Formulate this as a linear programming problem. SELECTED BIBLIOGRAPHY Bowman, E. H.: "Consistency and Optimality in Managerial Decision Making,” Management Scioxce, vol. 9, January 1963, pp. 310-321. : "Production Scheduling by the Transportation Method ef Linear Programming,” Operations Re- search, vol. 4, February 1956, pp. W0-103, Burrs, Exwoon 5.: Modern Production Management, Sth ed., New York: Wiley, 1977. Cottey, Joun L, Jn, Rosser LanDst, and Roser R. Fain: Production Operations Planning G Conthol, San Francisco: Holden-Day, 1977. Ewon, SAMUEL: “Five Approaches to Aggregate Production Planning,” AlIE Trensections, vol. 7, no. 2, Jane 1975, pp. 118-131. Fuuisn, Jack A.:""A Linear Programming Approach to Aggregate Scheduling," Academy of Management journal, vol. 18, no. 1, 1975, pp. 139-137. Ga arrit, JAY R.: "Solving Production Smoothing Problems," Management Science, vol. 15, no. 12, Au- gust 1969, pp. B6sS—B675. i: Haussuiarey, Feand 5. W. HEss: “A Linear Programming Approach to Production and Employment Scheduling.” Monsgenent Technelgy, vol 1, Jenusey 1960, pp, 45-52 ‘ Hout, Geta: Pluming Production tnvetrics end Wark Fore, Englewood Cli, N.J: Prentice-Hall, 1960 Franco Mopictiamt, and Heastar Simon: “A Linear Decision Rule for Production and Employ rent Scheduling" Management Siem, wol 2, ne. 1, Octobet 1955, pp. 1=30 1 sou Cons "Pam Proton Panne Meigen Scena 1390.1} 197. po} 843-866, / Kaajewsnt, L. J.. and H. E. Twomeson: “Efficient Employment Planning in Public Utilities,” The Bell Jounal of Economics end Management Science, Spring 1975. Lz, S. M.: “An Aggregative Resource Allocation Model for Hospital Administration,” Socio-Economic Planning Sciences, vol. 7, 1973, pp. 381-395. aiid L. J. Moone: “A Practice Approach to Production Scheduling, , Ist quarter, 1974, pp. 79-92. Let, W. B., and B. M. Kuusawata: "Simulation Testing of Aggregate Production Planning Models in an linplementation Methodology.” Managemen Science, vol 20, February 1974, pp. 903-911. Meuiciant, Josera, and Rosent Love: “Production Heuristics or the Aggregate Planaing Problem,” ‘Management Science, vol. 24, no. 12, August 1978, pp. 1242~1251.