Inventory systems for dependent demand

Transcription

1 Roberto Cigolini Department of Management, Economics and Industrial Engineering Politecnico di Milano 1

2 Overall view (taxonomy) Planning systems Push systems (needs based) (requirements based) Pull systems (stock based) Traditional Just in Time Dependent demand Independent demand The planning of requirements consists of the determination of What How much and When to order at every stage of the production process 2

3 Characteristics of pull (stock based) systems Objective: having always the required product stored in the warehouse (according to the service level) Required information: order issuing criteria (re order policy), i.e. the triggering mechanism Implicit hypotheses: Smoothed and even stock consumption Independent demands among finished products Reduced demand variance Distinctive feature: each phase of the production process only sees the warehouse immediately downstream and it is completely blind with reference to the remainder of the supply chain Physical flow Pull planning system Information flow Saw tooth profile over time Safety stocks based on variance Service level taken from the Gauss function This does not protect the inventory system against the so called bullwhip effect 3

4 A note on the bullwhip (Forrester, 1961) effect Even a very small change at the finished product level (end customer) may represent a remarkable source of variance when going upstream along the supply chain and/or along the bill of materials A bullwhip Distribution stage 40 Production stage Demand Variance (%) Components stage Example referred to the US automotive industry (see Anderson et al. 2000, Upstream volatility in the supply chain: the machine tool industry as a case study, Production & Operations Management, 9, (3), ) 4

5 A note on the bullwhip (Forrester, 1961) effect Inventory level Manufacturing lead time Finished products Re order point Re order Inventory level r 1 r 2 r 3 time Components Inventory level r 1 r 2 r 3 time Raw materials time r 2 5

6 Characteristics of push (needs / requirements based) systems Objective: calculating which, how many and when components, sub assemblies, parts, raw materials etc. are required to put a plan into operation i.e. to ensure that the customers orders due dates (deadlines) are respected Physical flow Pull planning system Information flow Physical flow Push planning system Information flow Required information: It is much greater than under pull systems, as it is needed to know the master production schedule, the bills of materials and to consider at the same time all the data referred to all the products and departments involved BOMs push MPS 6

7 Characteristics of push (needs / requirements based) systems Remarks: Requirements of components directly depend on a plan (e.g. the master production schedule) Requirements of components are therefore calculated and not estimated (i.e. derived from statistical analyses) as under pull systems In the end, the objective lies in coordinating the production dates (rendezvous) of components to manufacture finished products (or higher level components in the bill of materials) MRP means Materials Requirements Planning and it represents the procedure that implement the data processing needed by the push approach to manage inventories Production date 7

8 Characteristics of push (needs / requirements based) systems + MRP procedure operates in a recurrent way according to the so called 3S approach 1. Sum the requirements of the same component coming from different orders and referred to the same period 2. Split the overall requirements per period of each component according to the lot sizing policy 3. Shift backward over time the lot sized requirements according to the lead times reported in the bills of materials (to take into account the production routings) This leads to a plan of purchasing and manufacturing ordering proposals This plan in turn generates gross requirements of lower level components of the bill of materials The recurrent procedure is finished when the raw materials (i.e. the leaves of the BoM) are reached Explosion of the bills of materials 8

9 Example of MRP running (1/10) Finished products (top items) Motorcycle Lead time 3 weeks Re order policy Fixed EOQ, 150 pieces Initial inventory 200 pieces Reserved stock 30 pieces (1st week) Orders in progress 48 pieces (3rd week) Safety stocks 50 pieces Scrap rate 10 % Data referred to the chassis and known in advance Components Chassis Power Train Raw materials (leaves) 2 1 Front & rear wheels Frame Exhaust Gear Engine 9

10 Example of MRP running (2/10) Week Gross internal requirements (chassis) Gross external requirements (chassis) Before entering the first S (sum) you should know also the profile (over time) of the gross requirements (i.e. the demand) of your item. Gross requirements are divided into: 1. Internal requirements, i.e. originated from other finished products (e.g. a different kind of motorcycle) 2. External requirements, i.e. originated form customers (e.g. a chassis sold as spare part) 10

11 Example of MRP running (3/10) Week Gross internal requirements (chassis) Gross external requirements (chassis) = Gross total requirements (chassis) Now we enter the first S sum: The gross total requirements of the chassis are give by the sum period by period of the gross internal requirements and the gross external requirements 11

12 Example of MRP running (4/10) Week Gross internal requirements (chassis) Gross external requirements (chassis) Gross total requirements (chassis) Initial availability (chassis) = 120 = , i.e.: Initial _ inventory Reserved _ safety stocks stocks The gross total requirements consume the initial availability. So, at the beginning of period 2, initial availability accounts for 60, i.e At the beginning of period 3, initial availability accounts for 20, i.e At the beginning of period 4 initial availability is 0, given that 30 > 20, i.e. the gross total requirements exceed the initial availability 12

13 Example of MRP running (5/10) Week Gross internal requirements (chassis) Gross external requirements (chassis) Gross total requirements (chassis) Initial availability (chassis) Requirements (chassis) Once the initial availability is finished, all the gross requirements are converted into requirements 13

14 Example of MRP running (6/10) Week Gross internal requirements (chassis) Gross external requirements (chassis) Gross total requirements (chassis) Initial availability (chassis) Requirements (chassis) 10 x Scraps adjusted requirements (chassis) Now you have to take into account the 10% scrap rate So, all the requirements are to be multiplied by

15 Example of MRP running (7/10) Week Gross internal requirements (chassis) Gross external requirements (chassis) Gross total requirements (chassis) Initial availability (chassis) Requirements (chassis) Scraps adjusted requirements (chassis) Orders in progress (chassis) 48 Now you have to take into account the orders in progress, i.e. 48 pieces arriving at the beginning of period 3. They represent some sort of additional availability, apart from that they are potentially subject to scraps, while availability is not 15

16 Example of MRP running (8/10) Week Gross internal requirements (chassis) Gross external requirements (chassis) Gross total requirements (chassis) Initial availability (chassis) Requirements (chassis) Scraps adjusted requirements (chassis) Orders in progress (chassis) 48 The scraps adjusted requirements consume the orders in progress So, referring to period 3, the net requirements are 0, given that 48 > 11 (and 37 pieces are left) Referring to period 4, the net requirements are 0 again, given that > 22 (and 15 pieces are left) Referring to period 5, the net requirements are 40, i.e Net requirements (chassis) left 15 left = Once the orders in progress are finished, all the requirements are converted into net requirements 16

17 Example of MRP running (9/10) Week Gross internal requirements (chassis) Gross external requirements (chassis) Gross total requirements (chassis) Initial availability (chassis) Now you have to split (S2) net requirements by economic lots (of 150 pieces) Requirements (chassis) Scraps adjusted requirements (chassis) Orders in progress (chassis) = 62 left = 29 left Net requirements (chassis) Lot sized requirements (chassis) = 110 left = 95 left = 22 left 17

18 Example of MRP running (10/10) Week Gross internal requirements (chassis) Gross external requirements (chassis) Gross total requirements (chassis) Initial availability (chassis) Requirements (chassis) Scraps adjusted requirements (chassis) Orders in progress (chassis) 48 Finally you have to shift backward over time (S3) lot sized requirements by the lead time (3 periods) Net requirements (chassis) Lot sized requirements (chassis) Orders to issue (chassis) Lead time = 3 18

19 The recurrent approach of MRP Now you have on hand the plan of orders for the chassis It has to be converted into gross requirements of the wheels and of the frame, i.e. of the lower level components (raw materials in the example) Week Orders to issue (chassis) Gross internal requirements (wheels ) coming from the chassis x factor comes from the coefficient of use These gross internal requirements of wheels coming from the chassis maybe have to be summed up with gross internal requirements coming from other products and with gross external requirements (of the wheels) if they are sold as spare parts 19

20 Limits of MRP procedures Three major areas are referred to as critical system design features (Orlicky, 1974) 1. Production capacity is infinite While all the production systems are capacity constrained 2. Lead times are fixed and pre determined (a priori, outside MRP) While lead times result form the planning activity 3. Data used by MRP require a huge amount of space And MRP is subject to the garbage in garbage out (GIGO) law 20

21 Limits of MRP procedures 1. Production capacity is infinite MRP basically operates at infinite capacity, so the load of work centers is optimized only indirectly, through lot sizing rules. 600 Orders to issue MRP procedure by itself is unable to manage the peaks of workload 130 Production capacity Periods 10 21

22 Limits of MRP procedures 1. Production capacity is infinite Production capacity is managed through finite capacity post processors which are often based on (complex) linear programming (LP) and/or theory of constraints (ToC) approaches Step 1: initial infeasibility Step 2: Overloaded is either avoided or reduced by shifting orders backward Step 3: Overloaded is removed by shifting orders forward

23 Limits of MRP procedures 2. Lead times are fixed and pre determined (a priori, outside MRP) Lead times are used as input variables and they are not considered as function of (dependent on) the work load No protection is assured against the orders in the past phenomenon i.e. the orders that fall in periods prior to period 1, i.e. orders that would have had to be issued in the past (yesterday, last week / month etc.) Go back to the example of MRP running and suppose that the lead time of the chassis is 5 weeks instead of 3 week Net requirements (chassis) Lead time = 5 Lot sized requirements (chassis) Orders to issue (chassis) Order in the past 23

24 Limits of MRP procedures 2. Lead times are fixed and pre determined (a priori, outside MRP) MRP uses lead times as input, while their actual value is an output Lead times are variable by nature, due to technology and organizationrelated factors No protection is assured against the orders in the past phenomenon i.e. the orders that fall in periods prior to period 1, i.e. orders that would have had to be issued in the past (yesterday, last week / month etc.) In the end, lead times estimation is critical: Underestimating lead times leads to stock out (of components) and therefore it puts the entire logic of the production dates in crisis Overestimating lead times causes the planning horizon expansion, which implies: A lower data reliability, since in the long term the portfolio will be composed of forecasts and fewer certain orders An increase of components stock holding costs as they are manufactured longer before the time they are actually needed 24

25 Limits of MRP procedures 2. Lead times are fixed and pre determined (a priori, outside MRP) Lead times can be managed by shortening time buckets E.g. by using days as buckets instead of weeks Shortening time buckets requires (more) accurate forecasts and (more) frequent planning Finished product LT = 7 days, i.e. 2 weeks Components Example: Finished product Components Raw materials Week 1 Week 2 M T W T F Planning under weekly time bucketing Planning under daily time bucketing Past Future Week " 1" Week 1 Week 2 Week 3 Week 4 M T W T F M T W T F M T W T F M T W T F M T LT = 8 days, i.e. 2 weeks Raw materials A requirement scheduled for week 4 (if not absorbed by either availability or orders in progress) gives rise to an order in the past at the raw materials stage under a weekly bucketing approach, while it gives rise to an order for today under a daily bucketing approach 25

26 Limits of MRP procedures 3. Data used by MRP require a huge amount of space The volume of data is relevant, mainly for the bills of materials (BOMs) Consider Taurus tractor, equipped with different devices that correspond to various configurations of the same basic product To fully represent all the available alternatives for this (very simplified) example the (huge) number 3 x 2 x 5 x 2 x 2 x 3 x 3 = 1,080 of bills of materials required is: Feature Alternatives Description Number of wheels 3 4 (2 motion); 4 motion; 1 rear & 2 front Fuel engine 2 Petrol; diesel Power 5 20, 60, 80, 120, 200 kw Gears 2 Normal; normal & overdrive Steering 2 Normal; power steering Rear tow hook 3 Normal; strengthened; special Power takeoff 3 Absent; normal; special 26

27 Limits of MRP procedures 3. Data used by MRP require a huge amount of space To save space you can have to resort to super bills (product configurators) Through the analysis of commonalities, the bills of materials are arranged in modules (also called modular bills) Each module is a fictitious (artificial) bills that contains all the codes of one single option Taurus tractor Common to all Wheels Engine Power Gearbox Steering Tow hook Takeoff 4, 2 motion 4 motion 1 rear, 2 front petrol diesel 20 kw 60 kw 80 kw 120 kw 200 kw normal overdrive normal power steering normal special strengthened absent normal special Specific n 27

28 Limits of MRP procedures 3. Data used by MRP require a huge amount of space Strengths of super bills 1. They reduce the required space by 1/100 ore even more E.g. in the example of the Taurus tractor the number of bills of materials required is: n = 21 + n Which is anyway much less than: 3 x 2 x 5 x 2 x 2 x 3 x 3 = 1,080, i.e. the number of BoMs without super bills 2. They remarkably help data maintenance and to keep data consistency Power Taurus tractor Gearbox Steering normal overdrive You can easily add or remove features without having to resort to multiple data operations Taurus tractor Gearbox normal Taurus tractor Gearbox normal overdrive special overdrive 28

29 Limits of MRP procedures 3. Data used by MRP require a huge amount of space Strengths of super bills 3. They dramatically improve accuracy of the forecasting process The sales forecasts over the medium long term, usually calculated only by product type (e.g. Taurus tractor) are converted into (equally) reliable forecasts at the components / single option (part number) level This allows a longer Master Production Schedule (MPS) horizon Taurus tractor The coefficients of use (expressed as %) represent how popular is the addressed product within the sales mix Common to all Through the coefficients of use the forecast for Taurus tractor is translated (with the same reliability) into corresponding forecasts for each component Wheels Engine , 2 motion 4 motion 1 rear, 2 front petrol diesel Power kw 60 kw 80 kw 120 kw 200 kw 29

30 Executive summary Materials Requirement Planning (MRP) systems make the explosion of the bills of materials, by calculating (not estimating) which, how many and when components, sub assemblies, parts, raw materials etc. are required to ensure that the customers orders due dates (deadlines) are respected In the end MRPs coordinate the production dates (rendezvous) of components to manufacture finished products (or higher level components in the BoM) MRP is very useful to protect the inventory system against the bullwhip effect However, three major areas are referred to as critical system design features of MRP: MRP basically operates at infinite capacity Lead times are assumed as fixed and pre determined MRP requires a high volume of data 30

31 Further (suggested) readings Masuchun, W., Davis, S., Patterson, J.W. (2004) Comparison of push and pull control strategies for supply network management ina make to stock environment, International Journal of Production Research, Vol. 42, No. 20, pp Butman, J. (2002) A pain in the (supply) chain, Harvard Business Review, May, pp Caridi, M., Cigolini, R. (2002) Improving materials management effectiveness: a step towards agile enterprise, International Journal of Physical Distribution and Logistics Management, Vol. 32, No. 7, pp Koh, S.C.L., Saad, S.M., Jones, M.H. (2002) Uncertainty under MRP planned manufacture: review and categorization, International Journal of Production Research, Vol. 40, No. 10, pp Barzizza, R., Caridi, M., Cigolini, R. (2001) Engineering change: a theoretical assessment and a case study, Production Planning and Control, Vol. 12, No. 7, pp

32 Practice 1 Company Alpha manufactures product Beta, made up from 2 critical components (B and C), according to the represented relevant data, while the table reports the gross total requirements of Beta 2 Beta 3 Lead time Re order policy Initial inventory 2 periods Fixed EOQ, 250 pieces 350 pieces B 10% scrap rate 20% scrap rate C Reserved stock Orders in progress 100 pieces (1st period) 100 pieces (3rd period) Safety stock 50 pieces Scrap rate 5 % Period Gross total requirements You are required to calculate the gross requirements of B and C coming from Beta 32

33 Practice 1 short discussion Period Gross total requirements Initial availability Scraps-adjusted requirements Orders in progress 100 Net requirements Lot-sized requirements Orders to issue (of Beta) Gross requirements of B Gross requirements of C

34 Practice 2 Consider the represented bill of materials and the relevant data reported in the tables together with the gross total requirements of A B A D C E Code Coefficient of use Lead time (weeks) Initial inventory (pieces) Safety stock (pieces) Re-order policy A L4L B L4L C EOQ = 50 D L4L E EOQ = 200 Week Gross total requirements of A You are required to calculate the plan of orders to issue for all the involved components (i.e. B, C, D, E) 34

35 Practice 2 short discussion Period Gross requirements (of A) Initial availability (of A) Scraps-adjusted requirements (of A) Lot-sized requirements (of A) Orders to issue (of A) Gross requirements (of B) Gross requirements (of C) Initial availability (of B) Scraps-adjusted requirements (of B) Lot-sized requirements (of B) Orders to issue (of B) Gross requirements (of C) Initial availability (of C) Scraps-adjusted requirements (of C) Net requirements (of C) Lot-sized requirements (of C) Orders to issue (of C) Gross requirements (of D) Gross requirements (of E) Initial availability (of D) Scraps-adjusted requirements (of D) Lot-sized requirements (of D) Orders to issue (of D) Gross requirements (of E) Initial availability (of E) Scraps-adjusted requirements (of E) Lot-sized requirements (of E) Orders to issue (of E) Order in the past, manageable e.g. through safety stock 35