SLAM - A Case Sudy in Applying Lean to Job Shops

Submitted for inclusion in the proceedings of the 2005 ASCE Construction Research Congress, San Diego, CA, April 5-7, 2005.

SLAM – A CASE STUDY IN APPLYING LEAN TO JOB SHOPS Downloaded from ascelibrary.org by UNIVERSITY OF CALIFORNIA LIB. on 01/06/14. Copyright ASCE. For personal use only; all rights reserved.

Todd Brink and Glenn Ballard1 ABSTRACT Lean production has been applied broadly within manufacturing, but mostly to production systems that can be structured as flow lines. Can lean be applied to production systems that produce small numbers of a wide variety of custom products and thus do not lend themselves to a flow line structure; i.e., to job shops? Systems structured as job shops belong at the same end of the spectrum of production system types as do projects, so finding how to take job shops lean could have important implications for both manufacturing job shops and the application of lean to project based production systems such as those producing capital facilities, developing consumer products, and engineering software. Spancrete, a Midwest fabricator of concrete products, is achieving great success with its lean implementation. The primary tool is value stream mapping, which is done by employee teams with participants drawn from every level in the organizational hierarchy. Substantial performance improvements have been made, but arguably the most important change is cultural, the transformation of Spancrete into a learning organization. This paper reports the methodology and the results Spancrete has achieved and suggests how their methodology might be applied by others. KEY WORDS Concrete, fabrication, fabricator, implementation, job shop, lean, lean production INTRODUCTION Spancrete is a manufacturer of precast concrete products based in Waukesha, Wisconsin, with five precast fabrication plants, two concrete pipe plants, and one machinery plant in various locations in Wisconsin and Illinois. Spancrete employees more than 500 people and produces both custom and standard products, and thus its different production facilities can be characterized as job shops or as batch flow shops (more on this in the next section). In early 2003, in response to declining profit margins, Spancrete began implementing SLAM, Spancrete’s Lean Approach to Manufacturing2. 1

2

Todd Brink is the SLAM (Spancrete’s Lean Approach to Manufacturing) coordinator for Spancrete, a producer of precast and other concrete products. Glenn Ballard is the Research Director for the Lean Construction Institute, a principal of Strategic Project Solutions, and Associate Adjunct Professor at the University of California, Berkeley. [email protected] The term “lean” is used in different ways by different speakers. Some apply the term to narrower domains, such as manufacturing, while others have been inspired by Toyota’s accomplishments in both factory production and product development to seek a more fundamental theory of production in which the project is understood as the primary form or type of production (Koskela, 2000; Ballard, 2005). In

1 Copyright ASCE 2005

Construction Research Congress 2005 Construction Research Congress 2005

Downloaded from ascelibrary.org by UNIVERSITY OF CALIFORNIA LIB. on 01/06/14. Copyright ASCE. For personal use only; all rights reserved.


Among the many challenges of taking a company lean, Spancrete struggled with how to apply the models and lessons from Toyota to their own situation. How can rules and tools developed for assembly lines and worker-paced flow lines be applied to Spancrete’s production-to-order of standard and custom products? They decided to rely primarily on value stream mapping3, which has proven effective not only in achieving substantial improvements in the way work is done, but even more important, is causing a cultural revolution, transforming Spancrete into a learning organization (Senge, 1990). The first section below is on the types of production systems and the challenges they pose for implementation of lean concepts and techniques. That is followed by a description of Spancrete’s implementation methodology, then by a job shop example, a batch flow shop example, a summary of achievements and benefits from SLAM, and finally conclusions and generalization to other fabricators serving the construction industry and to construction projects themselves4. PRODUCTION SYSTEM TYPES AND CHALLENGES Definitions of production system types have their roots in Rutherford and Hayes’ 1979 Harvard Business Review articles (Rutherford & Hayes, 1979a and 1979b), in which they organized types in terms of both product and process characteristics. Roger Schmenner (Schmenner, 1993) adds projects to the Rutherford & Hayes typology, locating them on one extreme, where products are unique and flows are “very jumbled” (see Figure 1). Job shops are just next to projects because they produce custom products, typically in small lots, and thus have products following different routings simultaneously in a single production system. Batch flow shops come next, the difference being that they produce standard products, either in response to customer orders, which makes them more like job shops, or for stocking inventories of finished goods. Most publications on lean manufacturing have dealt with flow lines. Indeed, it might be said that one key objective of the Toyota Production System (Ohno, 1988) is to transform batch flows into line flows through such techniques as one piece flow and pull. Ballard, et al. (2003) describes one construction industry application of this approach, where the work in a precast concrete factory was restructured into production cells, each dedicated to a product type such as walls, beams, or columns. As long as the labor content and duration of products does not vary too greatly5, the production of these products can be organized as labor-paced line flows, with production cells periodically restructured to accommodate a changing product mix. One critical difference is in the this latter perspective, it may sometimes seem as if all things good about production and project management are claimed to be good because they are ‘lean’, but the truth is rather that a comprehensive and new theory is required by means of which to understand and to integrate together what is successful in practice, and to provide the framework for further advances in practice. “Lean” is the name for that now emerging theory. 3 Value stream mapping is the term used by Rother and Shook (1998) to describe what Toyota calls material and information flow diagrams, and are used to describe how material and information flow from receipt to delivery of a customer order. 4 Portions of this paper are based in Brink, 2004. 5 The problem is to balance the advantages of line flow (reduced cycle times, lower work-in-process inventory, greater flexibility, etc.) with the loss of labor capacity when there is a varying work load on the system. The greater the variation, the greater the loss.





nature of the products made in the job shop; either engineered-to-order or standard products, meaning customers choose from a catalog of product designs previously produced. The second critical difference is in the mix of products in production at any one time. Products may be engineered-to-order, but more or less similar in the steps taken to produce them and the time required to perform each step. Production cells (resources dedicated to a type of product, with production structured as a line flow) are feasible when there is a sufficient quantity of a type of product (e.g., shear walls), and when those products are sufficiently similar that the benefits of cell production exceed the loss of capacity resulting from differences in processing steps or durations from one product to the next. Process Pattern

Very jumbled flow; process segments loosely linked Jumbled flow; but a dominant flow exists

One of a Kind or Few

Low Volumes: many products

High Volumes: several major products

Very High Volumes: Standard product (Commodity) Scheduling; materials handling; shifting bottlenecks

Project Job Shop Batch Flow

Line flow: worker paced

Worker motivation; balance; maintaining enough flexibility

Line Flow

Line flow: machine paced

Line Flow

Continuous flow, automted and rigid flow, process segments tightly linked

Management Challenges

Management Challenges

Product Mix

Continuous Flow Bidding; delivery; product design flexibility

Quality (Product Differentiation); flexibility in output volumes

Capital expenses for big chunk capacity; tecnological change, materials mgmt; vertical integration

Price

Figure 1: Product/Process Matrix (from Schmenner, 1993)

To our knowledge, nothing is published on the application of lean techniques to job shops as such; i.e., when the job shop cannot be restructured as a line flow6. Factory Physics (Hopp & Spearman, 2000), perhaps the premier text on manufacturing management, uses batch flow systems on which to develop their ‘science of manufacturing’, naturally looking to make them more like line flow or continuous flow processes, and neglecting job shops altogether. Conner’s Lean Manufacturing for the Small Shop (Conner 2001) proposes to show how to make job shops lean, but assumes that products can be grouped into families sharing similar routings and assigned to production cells; i.e., assumes that they can be restructured as line flows. Duggan (2002) would seem to be explicitly about job shops, but Jeffrey Liker ends his foreword with “You must isolate a product family with some limits on the range of variation in cycle times or establishing flow will be hopeless.” (p.xii) 6

It should be noted that construction, one type of project based production system, has its own lean research and body of literature. See websites of the International Group for Lean Construction (www.vtt.fi/rte/lean) and the Lean Construction Institute (www.leanconstruction.org).





Spancrete has both job shops in which they produce their wet cast products (e.g., walls, beams, columns, tees, stairs), and batch flow shops, in which they produce dry cast (Spancrete hollowcore and wall panels) and concrete pipe products. Their job shops produce a wide range of products, arguably too great a range to allow organization into flow lines. However, their batch flow production is a candidate for transformation into or approximations of line flows, which is evident in the second example below: Crystal Lake Plank Production. The Spancrete story certainly includes triumphs in the application of lean concepts and techniques to their production processes, but an even bigger story is how they are becoming a learning organization (Senge, 1990) and the role played by their approach to lean implementation and their use of value stream mapping (Rother and Shook, 1998) in that transformation. APPROACH The guiding philosophy of Spancrete’s lean implementation is to provide top down vision and support, but to generate changes and improvements bottom up. Spancrete divided the entire company into manageable groups; e.g., Wet Cast Waukesha, Pipe Specialty Green Bay (see Table 1 for a more extensive list). They developed an implementation schedule,

Figure 2: Core Team with process map on wall

choosing to start with direct production in order to force upstream support groups to keep pace, trained everyone in the first group7, then selected a core team to be the instrument for improving processes within that group, consisting of everyone from executive level to the plant floor, picking laborers who seemed to ‘get it’ during training.

7

Training consists of a 3-4 hour class on basics of lean, 3 hours presentation and 1 hour simulation, taught by Spancrete personnel.





Core team formation takes 1-2 days, including training in value stream mapping (also done internally), after which the team members collect data for two weeks8, then meet to map the current state process in a long one day session. Soon thereafter, if no additional data collection is needed, the core team identifies areas of opportunity for improvement and develops a roughcut future state map of the process, plus an implementation plan. They meet again two weeks into the implementation plan to take stock and replan as needed. Every quarter, each core team revisits its processes and uses value stream mapping to make further improvements. TWO EXAMPLES: JOB SHOP AND BATCH FLOW To illustrate the SLAM implementation process and its impact, two examples are provided below, one for a job shop type production system and one for a batch flow type production system, Waukesha Wet Cast and Crystal Lake Plank Production, respectively. WAUKESHA WET CAST The first core team project was done in Waukesha Wet Cast, beginning in late February, 2003. Data collection revealed that laborers spent on average 3 hours a day looking for tools, and team members proposed to apply 5S9, including the use of tool shadow boards on the shop walls so tools would be readily available and more often returned to their proper location after use. The manager of Wet Cast resisted a 5S solution for fear that if tools were readily available, workers would steal them. After much argument and education, the manager had to be released. He simply was not able to embrace the new philosophy. His release opened the floodgates for worker enthusiasm and ideas for improvement. In short order, the core team cleaned up the work area, set up tool shadow boards (Figure 3), and organized cabinets for supplies with a person responsible for keeping them stocked (Figure 4). A helper who had only recently come to work for Spancrete made a naïve suggestion: ‘Instead of stringing extension cords all over the floor, which takes time and also poses a tripping hazard, why not drop down cords from the ceiling?’ That resulted in reel stations for electrical power cords routed over ceilings and walls, soon followed by compressed air, oxygen and acetylene, the gases used in cutting torches. Job shop versus flow shop at Waukesha Wet Cast Waukesha Wet Cast has 10 form beds serviced by a concrete batch plant, with concrete delivered to the form beds by two small tractors with 2.5 cubic yard buckets. Beds, batcher and tractors constitute the machines in the production system. Some form beds are made of wood and some from steel. They are different maximum sizes. Some forms 8

The standard guide to value stream mapping, Rother & Shook’s Learning to See (1998), advocates following products as they move through the production system, but that is clearly a best fit for flow lines, not job shops, with their jumbled flow and multiple product routings. Data collection is consequently a more difficult and time consuming task for mapping job shop processes. 9 “A checklist for good housekeeping to achieve greater order, efficiency and discipline in the workplace. It is derived from the Japanese words seiri, seiton, seiso, seiketsu, and shituke and adopted to the English equivalents of sort, straighten, scrub, systematize, and standardize.” P.xxiv, Imai (1997).





are flat and some are vertical (for cylindrical columns). Some are made for rectangular columns, some for beams, some for decks. The type and size of products produced in a single day vary widely, making it uneconomical to dedicate labor resources to production cells/flow shops. Even so, Spancrete has demonstrated the applicability and benefits of lean techniques to job shops. Techniques successfully applied include value stream mapping, pull mechanisms10 applied to processing, reductions in batch sizes11, 5S, raw material inventory control with kanbans, point of use storage (e.g., supply cabinets, drop down cords), reduction in changeover time12, and total productive maintenance13 (TPM).

Figure 3: Tool Shadow Board in Waukesha Wet Cast after 5S

Waukesha Wet Cast Results 5S implementation was just the start of a stream of innovations in Waukesha Wet Cast. As of July, 2004, their cost per cubic foot of product was down 27% from the February, 2003 baseline, labor productivity was up 67%. SLAM was off to a good start. EXAMPLE 2: CRYSTAL LAKE PLANK PRODUCTION The Crystal Lake Plant is devoted to production of Spancrete plank, a drycast product made in a highly mechanized way, using a single machine and six form beds. Planks are made in thicknesses of 6, 8, 10 or 12 inches, and are cut to length to suit customer needs. This is an example of a type of product the production of which can be restructured from batch flow into a worker-paced line flow. 10

11

12

13

“Initiating the delivery of inputs based on the readiness of the process into which they will enter to be transformed into outputs”, Glossary, Lean Construction Institute, www.leanconstruction.org A process batch is the amount of material or units of product processed together. This may occur simultaneously as in batching concrete or sequentially as when fabricators prefer making multiple units of product of one type before changing over to another type. A transfer batch is the amount of material or units of product moved together from a supplier to a customer workstation. The time required to change from producing one type of product or performing one type of activity to producing/performing another; also called setup time. TPM is the application of quality management to equipment, seeking to maximize equipment efficiency over its lifetime (cf. p. 9, Imai, 1997).





Prior to the core team’s work, plank was stack casted, which is industry standard practice. In other words, the next plank was cast on top of a previous plank in a form bed—a classic instance of producing in large batches. The core team decided to try single casting, which increased daily production from 5 beds to 6 with no increase in work time. Another benefit: stack casting forces you to make what you need last first and vice-versa, since you are burying product. Moving to single casting has made scheduling easier and more flexible—the usual result of reducing cycle time for producing a single unit of product. Crystal Lake Results The restructuring of the production process into one piece flow yielded an 18% increase in labor productivity (from .011 to .009 square feet of plank per direct labor hour); exactly what would be expected from increasing production output in the ratio 6:5 and not changing direct labor hours input. The operation was so mechanized and the number so small Spancrete thought they could never improve on it, but they did. They are now launching an initiative to reduce the changeover time from one thickness to another. WHERE SLAM IS NOW AND WHAT’S NEXT SLAM implementation is being done by work group as shown in Table 1. Table I: SLAM Work Groups Work Group Waukesha Wet Cast

Status Launched (now on 6th improvement cycle)

Results 27% cost reduction; 67% increase in productivity

Valders Wet Cast

Launched (on 5th cycle)

Crystal Lake Plank

Launched (on 4th cycle)

American Concrete Pipe Specialty

Launched (on 3rd cycle)

100% increase in tees, from pours every 2 days to daily pours 18% increase in productivity, greater flexibility to changes in demand 82% reduction in finished goods inventory

Drafting & Engineering Green Bay Wet Cast Valders Plank

Current

Waukesha Plank

2005

Corporate Office

2005

Jan 2005

Focus 5S in shop; TPM; cycle time reduction in shop from pull techniques, pouring concrete asap. Now working on yard operations 5S in shop; cycle time reduction. Current goal: turn 2 beds of insulated plank/day. Changed from stack casting to single piece flow. Now working to reduce changeover time from 8-inch to 10-inch plank. 5S and pull techniques on outside operations; now implementing 5S on inside operations Admin 5S, info transfer from Sales, materials ordering Outside production; large bridge beams & TTs

2005




Submitted for inclusion in the proceedings of the 2005 ASCE Construction Research Congress, San Diego, CA, April 5-7, 2005. Erection

2005

Customers

2005

Improve shared processes

Four work groups have been launched and have gone through multiple process mapping cycles, resulting in improvements in productivity, cost, cycle time, and inventory. A variety of lean techniques have been applied, including value stream mapping, 5S (plant and administrative), pull mechanisms for advancing work from one ‘workstation’ to the next, raw material inventory control with kanban, visual workplace, standardizing work, total productive maintenance (TPM), 5S, single piece flow (process batch size of one), providing materials and tools at the point of use, and changeover time reduction. Drafting and Engineering is now in its first cycle, challenged to increase its capacity in order to keep pace with the production groups they serve. Three additional production groups are scheduled for launch in 2005, plus the corporate office and all erection groups. In addition, Spancrete plans to reach out to its customers, helping them learn how to apply lean concepts and techniques to their own operations and to shared operations. Among the many opportunities foreseen, perhaps the greatest is to help customers improve their demand reliability and reduce the frequency of late changes in required delivery dates. IMPACT ON THE COMPANY AS A WHOLE Despite its partial implementation and the inevitable difficulties such as recalcitrant managers and fixed habits, SLAM has already had an enormous impact on the entire company. Consider the improvements in production from 2003 to 2004: • • •

Throughput increased from 565,898 cu. ft. to 1,134,966 cu. ft. Direct labor hours per unit of output decreased from .174 to .162 Raw material inventory turns increased from 17.14 to 25.15

On the ‘soft’ side of the ledger, it is widely agreed that Spancrete is a better place to work. The change in culture is something a visitor can almost feel. Everyone seems to have a story to tell about how something was improved, from choosing to use two tractors to deliver concrete to form beds as a means of reducing cycle time and labor delays, to reducing material inventories in the warehouse, to revealing to management that for lack of an ‘expensive’ repair14, the yard crane had been unable to turn left for the last five years! Not all groups have done their initial value stream mapping exercises, but they are already learning from those who have. One example is Spancrete’s Erection group, which has imitated Waukesha Wet Cast’s practice of having the warehouse kit together the materials needed for each job, cutting 45 minutes from the time spent getting ready to go to work each day.

14

Analysis revealed that a simple, inexpensive modification to a hydraulic cable fixed a problem that workers had lived with and worked around for years.





CONCLUSION There are several lessons to be learned from Spancrete’s lean implementation, but first some caveats: 1) To repeat the caution expressed by many others many times, it is rarely feasible to imitate solutions developed by others15. Every company and situation is different, so others should not expect success from slavish, unthinking imitation of Spancrete’s example. 2) Naturally the successes have been emphasized in this account, but it’s not all been rosy. Change is hard. It requires clear vision and firm leadership. Failures are inevitable, so be prepared to learn from them. Some people will not be able to change and will have to be let go—a painful event in any case, but especially in a small company with intense personal loyalties. With those caveats in mind, we will consider what can be learned, first regarding the application of lean techniques to job shops and batch flow shops, then regarding the transformation of a company into a learning organization. JOB SHOPS, BATCH FLOW SHOPS AND LINE FLOWS Batch flow shops producing standard products can be restructured as line flows when there is sufficient demand for the products or product families to economically justify the investment in separate resources, which is typically the case else the batch flow shop would not have been established in the first place. This paper has described the results of restructuring a precast fabricator’s batch flow shop, Crystal Lake plank production, into a line flow. A fabricator’s operations typically include both batch flows and job shops, the latter producing custom products. Even though some job shops cannot be transformed into line flows16, this paper has described one example, Waukesha wetcast, how lean techniques can still be applied, with substantial improvements in operational metrics such as inventories, cycle times, production rates and productivity. CHANGING THE CULTURE No matter how much processes and performance are improved, the prize goes to those who can sustain continuous improvement—what Womack and Jones (1996) called “the pursuit of perfection.” Jeffrey Liker (2003) proposes fourteen principles of the Toyota way, the last of which is “Become a learning organization through relentless reflection and continuous improvement”. Changing the culture is the critical change needed to get everyone working on making improvements all the time, and to inculcate the habits and disciplines of learning. Spancrete appears to be making that transformation. The following are necessarily speculative, but seem to the authors to be probable causes of Spancrete’s success: 15

16

For example, see p. xii of the foreword to Liker (2003) by Gary Convis, President of Toyota Motor Manufacturing, Kentucky, where he says “…each organization must develop its own way of doing business.” Future research is needed to better understand the principles of production system design for job shops and projects. One possibility to explore is the frequency of production system restructuring. Job shops such as Waukesha WetCast might be understood as production cells (line flows) of very short duration. Similarly, the application of resources to a daily production plan could be understood as a structuring of a production system intended to have a short lifetime.





• •

• •

A leader of the lean implementation effort solely dedicated to that task. The strategy of providing top down vision and support, but making changes from bottom up. This is an effective way of empowering employees, but senior managers, including the facilitator of the implementation effort, must be able to put their egos aside and support changes even when they think they have a better idea. Value stream mapping, which is an effective tool for bottom-up identification of waste and opportunities for improvement. Teams as the organizational unit of change, drawn from all hierarchical levels and various functional departments. This promotes communication and learning across both horizontal and vertical levels within a company.

LEARNING FROM SPANCRETE’S EXAMPLE Bearing in mind the warning against unthinking imitation, nonetheless others can learn from Spancrete’s example. Other precast concrete fabricators come first to mind, especially those that provide engineered-to-order products and routinely have a broad mix of products in production. Indeed, job shops producing all types of products can beneficially apply the Spancrete approach, whether or not they serve the construction industry. Beyond that come projects as a type of production system designing and making a single, unique product. Granting that uniqueness, even so projects repeat processes, even when the products of those processes differ one from another; e.g., processes for pulling wire, placing concrete, reviewing submittals, evaluating change orders, planning production, selecting specialty contractors, and so on. Mapping those processes can reveal opportunities for eliminating waste, and better yet, can be a social instrument for engaging project members at every level in the improvement and learning process. REFERENCES Ballard, Glenn (2005). “Construction: One Type of Project Based Production System.”, to be presented at the SCRI Forum, Construction Beyond Lean, January 19, 2005, Salford University, Manchester, U.K. Ballard, Glenn, Nigel Harper, and Todd Zabelle (2003). “Learning to See Work Flow: Application of Lean Production Concepts to Precast Concrete Fabrication.” Journal of Engineering, Construction and Architectural Management, 10(1), Blackwell Publishers, Oxford, U.K. pp. 6-14. Brink, Todd (2004). “Lean Manufacturing: Spancrete’s Lean Implementation Plan and Results”. Proceedings of the 2004 Lean Construction Institute Congress, Park City, Utah, September, 2004. Available at www.leanconstruction.org. Conner, Gary (2001). Lean Manufacturing for the Small Shop. Society of Manufacturing Engineers, Dearborn, MI. 258 p. Duggan, Kevin J. (2002). Creating Mixed Model Value Streams: Practical Techniques for Building to Demand. Productivity Press, New York, NY. 206 p.





Hayes, Robert H. and Steven C. Wheelwright (1979a). “Link Manufacturing Process and Product Life Cycles”. Harvard Business Review, Harvard Business School, Boston, MA, January-February, 1979. Hayes, Robert H. and Steven C. Wheelwright (1979b). “The Dynamics of ProcessProduct Life Cycles”. Harvard Business Review, Harvard Business School, Boston, MA, March-April, 1979. Hopp, Wallace J. and Mark L. Spearman (2000). Factory Physics, 2nd ed. McGraw-Hill, New York, NY. 698 p. Imai, Masaaki (1997). Gemba Kaizen: A commonsense, low-cost approach to management. McGraw-Hill, New York, NY. 352 p. Koskela, Lauri (2000). An exploration towards a production theory and its application to construction. VTT Publications, Technical Research Centre of Finland, Espoo, Finland. 296 p. Liker, Jeffrey (2003). The Toyota Way: 14 Management Principles from the World’s Greatest Manufacturer. McGraw-Hill, New York, NY. 330 p. Ohno, Taiichi (1988). The Toyota Production System: Beyond Large-Scale Production.. Productivity Press, Portland, OR. 143 p. Rother, Mike and John Shook (1998). Learning to See: Value Stream Mapping to Create Value and Eliminate Muda. The Lean Enterprise Institute, Brookline, MA. 96 p. Schmenner, Roger W. (1993). Production/Operations Management, 5th ed. Prentice Hall, Englewood Cliffs, NJ. 825 p. Senge, Peter M. (1990). The Fifth Discipline: The Art & Practice of the Learning Organization. Doubleday/Currency, NY, NY. 424 p. Womack, James P. and Daniel T. Jones (1996). Lean Thinking: Banish waste and create wealth in your corporation. Simon & Schuster, New York, NY, 1996. 350 p.





