PUSH & PULL: Production Paradigms
Operations Management Dr. Mark P. Van Oyen
Physics should be explained as simply as possible, but no simpler. – Albert Einstein
Based on Chap’s 14 & 15 of Stevenson; Chap’s 11 & 14 of Gaither & Frazier. Filename: push-pull-lec.ppt
• Introduce two competing production control approaches/philosophies and critique them. • Gain insight into when to use each of these (or look for something else)
Blizzard of Buzzwords
• MRP: Materials Requirements Planning • MRP II: Manufacturing Resources Planning • ERP: Enterprise Resources Planning • TBC: Time-based Competition • JIT: Just in Time • OPT: Optimized Production Technology
PUSH – MRP, MRP II, part of ERP systems • KEY IDEA: If the finished product is due at time T,
• • • • when to begin assembly? when to begin subassembly? when to begin fabrication? when to procure raw materials?
• • •
Think about the process of preparing a Thanksgiving Day dinner: an explosion of parts traced backward in time from the start of dinner. Best applied to process-focused operations (associated with job shops, customized products and/or high product variety, make-to-stock production) MRP is a computer-based information system designed to handle ordering and scheduling of dependent-demand jobs/parts.
MRP Assembly Time Chart Backwards explosion of components to establish release times
Procurement of raw material D Fabrication of part E
Procurement of raw material F Procurement of part C Procurement of part H Procurement of raw material I Fabrication of part G Final assembly and inspection
Purpose of Production Control
Objective: Meet customer expectations with on-time delivery of
correct quantities of desired specification without excessive lead (cycle) times or large inventory levels.
Push Systems: MRP, MRP II
• General. Due-date driven planning hierarchy with detailed scheduling of work releases (Thanksgiving dinner analogy) • Control release rate; observe WIP level • Underlying model often inappropriate (dirty little secret).
Pull Systems: JIT (a.k.a. Kanban, but that’s a misnomer)
• Suitable only for repetitive or assembly line manufacturing; • Reduces congestion by authorizing work releases • Control WIP level; observe system throughput • Improves production capacity, quality • may bring high-stress to both workers and suppliers. 6
What are the concepts PUSH and PULL?
• Make what you decide to make and put it in the downstream inventory buffer without regard to what is happening downline. • Associated with make-to-stock production, inventories, and higher WIP levels • Frequently based upon due-dates quoted to customers if products require customization.
• Demands serve as triggers to start production • A workstation pulls jobs/output from the upstream (preceding) station as it is needed • A workstation produces in response to pull signals generated by usage/demand at downstream station(s) • Associated with low inventory levels & stock • “You don’t never make nothin’ and send it no place. Somebody has got to come and get it.” (see R. Hall.)
Push vs. Pull Production Systems
A Simple Production Scenario
Finished Goods Inventory (FGI) Buffer
Buffer Raw Materials
• Push system: Completed A jobs are sent (pushed) to B regardless of system conditions. • Pull system: Removal of a job from the finished goods buffer signals the execution of a B job, which draws from the buffer in front of B and signals the execution of an A job. Work is pulled from raw materials through the production line all the way to finished goods.
Operational Contrast: PUSH and PULL
PUSH (MRP) - work as much as
you can as fast as you can
Kanban Authorization Cards authorize production
When a stock item is removed, it’s Kanban card is sent back to beginning of cell to authorize production of a new job
Advantages of Pull
• Observability: we can see WIP but not capacity. • Efficiency: pull systems require less average WIP to attain same throughput as equivalent push system. • Robustness: pull systems are less sensitive to errors in WIP level than push systems are to errors in release rate. • Quality: pull systems require and promote improved quality.
Am. Prod. and Inven. Control Soc. (APICS) definition of PULL (JIT) A philosophy of manufacturing based on planned elimination of all waste and continuous improvement of productivity. … The primary elements of JIT are to have only the required inventory when needed; to improve quality to zero defects; to reduce lead times by reducing setup times, queue lengths, and lot sizes; to incrementally revise the operations themselves …. In the broad sense, it applies to all forms of manufacturing: job shop and process as well as repetitive.
What’s In a Name?
• “Today and Tomorrow” by Henry Ford in 1926 an approach surprisingly similar to JIT
• JIT = The Toyota Production System (by Taiichi Ohno in the 60’s)
• • • • • • IBM: Continuous flow manufacture HP: Stockless Production GE: Management by sight Boeing: Lean Manufacturing Motorola: Short Cycle Manufacturing GM: Synchronized production
JIT is really a philosophy with some PULL mechanisms at its core.
• Inventory (muda) is regarded as evil and every attempt is made to reduce inventory Tends to work best in repetitive (assembly line) operations which are product focused or at least based on cellular manufacturing Designed for level demand and a smooth production flow Both the processing and the movement of jobs/materials occur just as they are needed and usually in small lot sizes.
• Smooth production (level production planning) • Reduce time and cost of setups. • Increase production capacity through crosstraining and worker organization • Emphasize preventive maintenance. • Make high quality products, with high quality raw materials and processes. • Establish long-term supplier relationships
• stress cooperation and reliability, not price. • Multi-echelon supply chain with fewer “directly reporting” suppliers
JIT tends to be high stress for workers, managers, and suppliers
Summary JIT Goals and Building Blocks Stevenson Figure 15-3
Ultimate A Goal balanced rapid flow
Supporting Goals Reduce setup and lead times Eliminate waste Eliminate disruptions Make the system flexible Minimize inventories
Traditional Supplier Network
Stevenson Figure 15-4a Buyer
Supplier Supplier Supplier Supplier Supplier
Tiered Supplier Network
Stevenson Figure 15-4b
First Tier Supplier Second Tier Supplier
Third Tier Supplier
You Push, and I’ll Pull
PUSH Reality of MRP: • Given inputs: How much is desired and when (which can be interpreted as a target throughput (TH), but is phrased in terms of deadlines) • Controls Release Times (thus fixing ideal/planned cycle times (CT)) • Measures WIP and realized CT as actual system outputs PULL Reality: • Given inputs: target TH, explicit desire to keep WIP to a minimum • Controls WIP (by triggering/authorizing production at each station) • Measures TH and CT as actual system outputs 18
What’s so great about PULL?
• can attain the same throughput as PUSH with less WIP
Ease of Control
• Work release is based on WIP levels, which are easy to see; Whereas the release rates and release times in PUSH are not.
• If we use the wrong WIP level in a PULL policy, the (ill) effect on system performance is small compared to a comparable percent error in release rate or release time.
PRO: shorter queues and cycle times help to quickly detect and provide pressure for a quick solution CON: demands high quality
Support for improving quality
TBC = Time-based Competition (introduced H.B.R. 1988) was a successor to JIT. TBC is a SHIFT IN FOCUS: from costbased mgmt. To time-based mgmt. What’s all the fuss about TIME – Reducing Cycle Time? 1. Gaining market share. A company that can introduce new products quickly AND fulfill demands with a very short cycle time can carve out new market niches (which are typically characterized by product customization and very unpredictable demand) - e.g. Ingersoll cutting tools achieved a 600% increase in market share in only a few years, even though the market size did not grow in aggregate. 2. Supports E-commerce. E-commerce will demand high levels of agility. Christmas 1999 showed tremendous ecommerce enthusiasm by customers, but this turned to disappointment as many companies could not keep cycle times down. 20
3. Better responsiveness to the customer. Consumers are getting spoiled; they have come to expect short lead times and buy from companies that can deliver them. 4. Better internal quality with less scrap. By keeping cycle times short, the amount of WIP wasted between the occurrence of a process (systemic) defect/breakdown and regaining an in-specification, quality process is reduced. 5. Reduced reliance on forecasts. The need for forecasting (which is always inaccurate!) is reduced as cycle times shrink. If cycle times are below a critical threshold, then products can be “made-to-order”, which represents a truly lean, JIT system (e.g. Dell Computer) 6. More flexibility. If cycle times are short enough, one can commit raw material to a particular product at a later point in time, thereby allowing production to flex with evolving demand patterns. 7. Reduced WIP. Provided throughput is held constant, Little’s Law proves that WIP must drop as cycle time 21 shrinks.
Beyond JIT - the Future of PULL
OLD: JIT began at Toyota with the Kanban approach, NEW: refinements such as CONWIP and QRM. About QRM = Quick Response Manuf. (R. Suri, Productivity Press, 1996): • A new Material Planning & Control Approach that combines PUSH and PULL • A “system”, including a written strategy that everyone in the company buys into and can understand how to use it to move toward the QRM objectives. • Has relatively complex production control approach based on cellular manuf - called POLCA. 22
QRM Offers Critique of KANBAN Approach to JIT
Lean Thinking by Womack & Jones 1996 = JIT “title -defending champion” QRM by R. Suri 1996 = “usurping contender” ???,
which is best suited to very small batch-size production and companies with a very large number of products with highly variable demand for at least some of the products
1. Eliminate Muda (waste in Japanese), which is viewed as
WIP inventory. QRM: There are many forms of waste beside inventory, Even focusing on WIP, Kanban can be inefficient in some systems: • Kanban works well in applications like automotive which have customization only through combinations of predefined standardized products - but it’s poor for producing truly customized products
QRM Offers Critique of KANBAN Approach to JIT
2. Implement production flow based upon • Paced production line with “Takt” time defining the “pulse period” of the entire line. This requires careful development of standardized tasks and units of work. • Heijunka (Japanese) = the production schedule is LEVEL QRM: We don’t want level schedules, because that means the demand is flat and schedules can be frozen. We want fast response to growth. 3. Implement Kanban for each and every product type. QRM: Some companies have many different product types (thousands!) which can all be manufactured from one or several facilities (think of group technology and product families). This requires one set of Kanban cards for each type, which is overwhelming and translates into large inventory levels, even with small Kanban card counts.
Question: If pull is so great, why do MRP II (ERP) systems sell? Answer: Manufacturing involves planning as well as execution.
Planning Push Pull Good Bad
Execution Bad Good
Planning and doing are separate parts of the same job … There must be at least a trace of doing in one’s job. Otherwise one dreams rather than performs. Once cannot, above all, do only; without a trace of planning in his job, the worker does not have the control he needs even for the most mechanical and repetitive routine chore. - Peter Drucker, 1954 25
MRP II Planning Hierarchy
Demand Forecast Capacity, Resource Planning Rough-cut Capacity Planning Bills of Material Inventory Status Aggregate Production Planning Master Production Scheduling
Material Requirements Planning
Capacity Requirements Planning
Job Dispatching (sequencing)
Bliss Beyond the Buzzwords