Glenn Ballard, LookAhead Planning - The Missing Link in Production Control
Glenn Ballard, LookAhead Planning - The Missing Link in Production Control
Glenn Ballard, LookAhead Planning - The Missing Link in Production Control
T he Combinat ion of Last Planner Syst em and Locat ion-Based Management Syst em
Glenn Ballard
Int egrat ing design planning, schedule and cont rol wit h Deplan
Hyun Jeong Choo
© 1997 by Glenn Ballard. All Rights Reserved.
Technical Report No. 97-3, Construction Engineering and Management Program,
Civil and Environmental Engineering Department, University of California, Berkeley, CA 94720
Proceedings 5th Annual Conference of the International Group for Lean Construction,
Griffith University, Gold Coast, Australia. July, 1997.
LOOKAHEAD PLANNING:
THE MISSING LINK IN PRODUCTION CONTROL
by
Glenn Ballard1
Abstract: Lookahead planning is one of the decision functions that constitute production control
systems (Melles and Wamelink 1993). It stands between overall project coordination schedules
and short term crew-level commitments, screening out scheduled activities that “should” but
cannot be done and improving the success rate of completing the tasks assigned in weekly and
daily plans. When measured against such objectives, current industry lookahead planning is
poorly performed. A case study is presented to illustrate current procedures and performance,
and suggestions are offered for improvement.
Preface
This paper reports on one aspect of a larger research program devoted to the topic of
production control in construction. A differentiating characteristic of this research is its selection
of the production unit and the assignment as the unit of analysis. Another differentiating
characteristic is its conceptual framework, which posits planning as a process of reducing
uncertainty and maximizing throughput, counter-posing plan reliability to resource redundancy as
alternative strategies for managing in conditions of uncertain work flow. For prior publications
from this research program, see Ballard and Howell (1997).
Introduction
Most construction projects issue a ‘master’ schedule at or near the beginning of the
construction phase, extending from beginning to end of the project. Such schedules may serve
many purposes, from long term coordination to specifying terms of payment. However, such
initial, total project schedules cannot be accurately detailed too far into the future because of lack
of information about actual durations and deliveries. Consequently, most construction projects
use some form of short term schedule to coordinate and direct the various trades and crews
working on the job. These schedules are often called “lookahead schedules” because many look
ahead several weeks into the future. Practice varies widely regarding the extent of lookaheads,
their level of definition of activities relative to master schedules, the frequency of issue, their
update and use in learning from experience, etc.
Lookahead schedules are commonly used in the construction industry in order to focus
management attention on what is supposed to happen at some time in the future, and to
encourage actions in the present that cause that desired future. However, lookahead schedules
are rarely conceived as having the specific purpose of producing sound
assignments, nor are procedures provided for lookahead processes. Usually, a lookahead
schedule is simply a drop out from the higher level schedule, occasionally at a greater level of
detail, but with no screening of scheduled activities against soundness or other criteria. The
prevailing idea seems to be simply that thinking ahead is beneficial.
1
Glenn Ballard is a founding member of IGLC, a construction industry consultant, and a
Lecturer in the Construction Engineering & Management Program, Department of Civil &
Environmental Engineering, University of California at Berkeley. ballard@ce.berkeley.edu
1
© 1997 by Glenn Ballard. All Rights Reserved.
Technical Report No. 97-3, Construction Engineering and Management Program,
Civil and Environmental Engineering Department, University of California, Berkeley, CA 94720
Proceedings 5th Annual Conference of the International Group for Lean Construction,
Griffith University, Gold Coast, Australia. July, 1997.
In previous papers (Ballard and Howell 1994), lookahead planning has been
differentiated from commitment planning, which often takes the form of weekly work plans
produced for each crew or subcrew of each trade. Measurement of the PPC (percent plan
complete: number of completed assignments divided by the total number of assignments) of
weekly work plans has revealed a chronic and widespread problem of low plan reliability. This
problem is of vital importance because of its adverse impact on labor productivity both of the
production unit that has a low PPC and those production units that are their customers and which
inherit the uncertainty passed onto them. Project durations are also extended by low plan
reliability, because of the extended durations of installation activities and because of the time
and material buffers that the industry has come to rely on to compensate for uncertainty (Howell
and Ballard, 1995).
Lookahead planning is proposed as the key to improving PPC, and consequently the key
to reducing project cost and duration. In this paper, a case study is used to illustrate efforts to
improve PPC through the early restructuring of a project production control system.
2
© 1997 by Glenn Ballard. All Rights Reserved.
Technical Report No. 97-3, Construction Engineering and Management Program,
Civil and Environmental Engineering Department, University of California, Berkeley, CA 94720
Proceedings 5th Annual Conference of the International Group for Lean Construction,
Griffith University, Gold Coast, Australia. July, 1997.
1 WEEK PLAN
PROJECT: Pilot FOREMAN: PHILLIP
ACTIVITY DATE: 9/20/96
Est Act Mon Tu Wed Thurs Fri Sat Sun PPC REASON FOR VARIANCES
Gas/F.O. hangers O/H "K" xxxx xxxx 0% Owner stopped work
(48 hangers) Sylvano, Modesto, Terry (changing elevations)
Gas/F.O. risers to O/H "K" xxxx xxxx xxxx xxxx 0% Same as above-worked on
(3 risers) Sylvano, Mdesto, Terry backlog & boiler blowdown
36" cond water "K" 42' xxxx xxxx xxxx 100%
2-45 deg 1-90 deg Charlie, Rick, Ben
Chiller risers (2 chillers wk.) xxxx xxxx xxxx 20% Matl from shop rcvd late Thurs.
Charlie, Rick, Ben Grooved couplings shipped late.
Hang H/W O/H "J" (240'-14") xxxx xxxx xxxx xxxx xxxx xxxx 100%
Mark M., Mike
Cooling Tower 10" tie-ins (steel) xxxx xxxx xxxx xxxx xxxx xxxx 70% Some work in next week's
(2 towers per day) Steve, Chris, Mark W. sched. was included this week.
Weld out CHW pump headers xxxx xxxx xxxx xxxx xxxx xxxx 100%
"J" mezz. (18) Luke
Weld out cooling towers (12 towers) xxxx xxxx xxxx xxxx xxxx xxxx 60% Eye injury. Lost 2 days
Jeff welding time
F.R.P. tie-in to E.T. (9 towers) 50% xxxx xxxx xxxx xxxx xxxx xxxx 100%
Firt, Packy, Tom
WORKABLE BACKLOG
Boiler blowdown-gas vents
-rupture disks
Figure 2 is a later weekly work plan by the same front line supervisor showing the
addition of a column for “Make Ready Needs”, a response to the fact that front line supervisors
have responsibilities for making assignments ready to be carried out, even after they appear on
weekly work plans. Note that 3 of 3 assignments were 100% completed.
WORKABLE BACKLOG
Figure 3 shows the PPC for all the mechanical contractor’s crews from the week of
10/7/96 through the week of 1/20/97, when the project was entering the turnover stage. In that
3
© 1997 by Glenn Ballard. All Rights Reserved.
Technical Report No. 97-3, Construction Engineering and Management Program,
Civil and Environmental Engineering Department, University of California, Berkeley, CA 94720
Proceedings 5th Annual Conference of the International Group for Lean Construction,
Griffith University, Gold Coast, Australia. July, 1997.
period, average PPC improved by roughly 10%, from 55% to 65%. The project was extremely
successful, with a gross margin (operating profit) of approximately 30%. Some of that financial
performance was clearly a function of pricing, but company and project managers attribute a
large part to the improvement in plan reliability and have committed to extending the production
control system to all its projects.
80%
70%
60%
50%
PPC
40%
30%
20%
10%
0%
2-Dec
9-Dec
7-Oct
16-Dec
23-Dec
30-Dec
4-Nov
6-Jan
14-Oct
21-Oct
28-Oct
11-Nov
18-Nov
25-Nov
13-Jan
20-Jan
Figure 3: Pilot Project PPC with 4 Week Moving Average
One of the reasons why PPC improved is because reasons (see Figure 4) for failing to
complete assignments were tracked and attacked. 347 of 587 assignments were completed as
planned over the entire study period, for an average PPC of 59%.
80
70
60
Number of Mentions
50
40
30
20
10
0
Materials Prerequisite Change in Manpower Under Unidentified Rework Design Info Methods Weather Testing/
Work Priorities Estimated Inspection
delays
4
© 1997 by Glenn Ballard. All Rights Reserved.
Technical Report No. 97-3, Construction Engineering and Management Program,
Civil and Environmental Engineering Department, University of California, Berkeley, CA 94720
Proceedings 5th Annual Conference of the International Group for Lean Construction,
Griffith University, Gold Coast, Australia. July, 1997.
Of the 249 assignments that were not fully completed, 71 were because of late or
defective materials, 42 because prerequisite work was not completed, 37 because of changes in
priorities, 33 because of absenteeism or accident (manpower), 23 because of failure to
accurately estimate the amount of labor time required to execute assignments, and so on. Over
the course of the project, improvement was made in the areas of materials (Figure 5),
prerequisite work (Figure 6), manpower (Figure 7), and overplanning (Figure 8). These gains
were partially offset by increases in the number of changes in priorities later in the project
(Figure 9).
20
15
10
5
0
1
11
13
15
17
19
Weeks
Figure 5: Materials
20
10
0
1
11
13
15
17
19
Weeks
5
© 1997 by Glenn Ballard. All Rights Reserved.
Technical Report No. 97-3, Construction Engineering and Management Program,
Civil and Environmental Engineering Department, University of California, Berkeley, CA 94720
Proceedings 5th Annual Conference of the International Group for Lean Construction,
Griffith University, Gold Coast, Australia. July, 1997.
20
10
0
1
11
13
15
17
19
Weeks
Figure 7: Manpower
20
10
0
1
11
13
15
17
19
Weeks
20
10
0
1
11
13
15
17
19
Weeks
Weekly work plans were required not only of the construction crews, but also from
individual supervisors, managers, and engineers. Investigation of materials problems revealed
that project engineers responsible for expediting and other materials-related tasks were not
always on the same page with production. Project engineers began to present their weekly work
6
© 1997 by Glenn Ballard. All Rights Reserved.
Technical Report No. 97-3, Construction Engineering and Management Program,
Civil and Environmental Engineering Department, University of California, Berkeley, CA 94720
Proceedings 5th Annual Conference of the International Group for Lean Construction,
Griffith University, Gold Coast, Australia. July, 1997.
plans for review in staff meetings (which were also instituted as a regular weekly event) and to
post them on their office doors. Coordination improved as a result. Project engineers reported
that they were relieved to have a mechanism for knowing that they were doing the right thing.
The material handling supervisor was equally excited to have a way of identifying and prioritizing
his tasks. He received and reviewed each superintendent’s lookahead schedule and each
foreman’s weekly work plan, interviewed each of them to determine exactly what he needed to
get done and when, then had them sign off on his detailed plan for each day of the following,
showing what material is to be received, using what equipment, what material is to be moved
into staging areas, etc.
Additional actions on reasons included: meeting with the fabrication shop to improve
deliveries, communicating lookahead schedules and weekly work plans to other contractors each
week, and asking workers to give a week’s notice of non-emergency absences.
The mechanical contractor had a centralized detailing crew and fabrication shops at its
home office. Planning must obviously extend beyond one week ahead in order to coordinate with
such internal suppliers, with external suppliers, as well as with other trades. Figure 10 shows a
typical 5 Week Lookahead, so-called because it extends five weeks into the future, although next
week is not shown. Some of its relevant features are the identification of assignments by front
line supervisor and crew, the scheduling of those assignments, and the identification of actions
needed to make the assignments sound as regards materials, design, or prerequisite work.
ACTIVITY 1/13/97
1/13/97 1/20/97
# 1/27/97
# 2/3/97
# NEEDS
M T W T F S M T W T F S M T W T F S M T W T F S
Scott's crew
"CUP" AHUs-10 CHW, 2 HW X X X X X X X X X X X X X X X CHW delivers 1-8-97 thru
1-13.HW delivers 1-20.
Punch, label, & tag AHUs x x x Materials on site
Ron's crew
DI Steam to Humidifier x x x Materials on site
7
© 1997 by Glenn Ballard. All Rights Reserved.
Technical Report No. 97-3, Construction Engineering and Management Program,
Civil and Environmental Engineering Department, University of California, Berkeley, CA 94720
Proceedings 5th Annual Conference of the International Group for Lean Construction,
Griffith University, Gold Coast, Australia. July, 1997.
1) Shape work flow in the best achievable sequence and rate for achieving project objectives
that are within the power of the organization at each point in time.
2) Match labor and related resources to work flow.
3) Produce and maintain a backlog of assignments for each frontline supervisor and crew,
screened for design, materials, and completion of prerequisite work at the CPM level.
4) Group together work that is highly interdependent, so the work method can be planned for
the whole operation.
5) Identify operations to be planned jointly by multiple trades.
The steps involved in the process each week are presented in Figure 11:
UPDATE AND ALLOW ONLY APPLY SCREEN CPM TRANSLATE MATCH LIST ACTIONS
ADJUST SOUND QUALITY SCHEDULED SCHEDULE WORKLOAD NEEDED TO
MASTER ASSIGNMENTS CRITERIA TO ACTIVITIES ACTIVITIES INTO AND MAKE
SCHEDULE INTO WEEKS ASSIGNMENT BEFORE ASSIGNMENTS CAPACITY ASSIGNMENTS
ONE AND TWO. S BEFORE ENTRY INTO READY WHEN
ADVANCING LOOKAHEAD SCHEDULED.
BY WEEK WEEKS
Step 1: Enter the latest status and forecast information into the project master schedule. Adjust
starts, completions, sequences, and durations accordingly.
Step 2: Do not allow any assignments into week one that are not ready, except by project
management decision. Ask the foreman if each assignment can be completed in week
one, recognizing that he/she may have to determine completion of prerequisite work at
the item level, arrange for prework such as scaffolding, and coordinate the use of shared
resources such as equipment or special tools. Allow that amount of work into week one
that can be completed in the week.
Step 3: Examine the remaining weeks in the lookahead, except for the last, moving from present
to future. Screen out any assignments that cannot be made ready when scheduled. Try
to maintain for each crew an amount of assignments twice that which can be completed
in a week.
Step 4: Identify those activities scheduled to start or complete in the lookahead week (e.g. Week
5 for the mechanical contractor) and screen out any activities that you do not know can
be made ready to assign when scheduled. Take into consideration the status of design,
including pending changes or open issues, the availability of materials and components
needed for each activity, and the probability that prerequisite work will be complete when
needed.
Step 5: Translate lookahead week activities into the language of assignments, grouping highly
interdependent operations that should be planned as a whole, and identifying operations
to be planned jointly by multiple trades.2
2
Different trades have different primary installation units. Piping and plumbing use isometrics, while electricians talk in terms
of cable pulls and junction boxes. Each trade should have a list of the work within its scope expressed in terms of its own
installation units in order to determine material and labor requirements. A convenient tool is a database of such units, with fields
allowing quantifying, assigning and tracking. Properly structured, the database makes work flow visible and allows for it to be
managed.
8
© 1997 by Glenn Ballard. All Rights Reserved.
Technical Report No. 97-3, Construction Engineering and Management Program,
Civil and Environmental Engineering Department, University of California, Berkeley, CA 94720
Proceedings 5th Annual Conference of the International Group for Lean Construction,
Griffith University, Gold Coast, Australia. July, 1997.
Step 6: Calculate the earnable manhours or otherwise quantify the labor content of the work in
the lookahead week. If that amount of work falls below the amount needed to maintain
schedule and if you will have the labor capacity to do that amount of work, advance work
from the master schedule to the extent practical. If the resultant amount of work falls
below the current work force, reduce the work force, or decide how to use the excess
labor time. If that amount of work exceeds the current or projected work force, decide
whether or not to increase labor to accelerate progress.
Step 7: Produce a list of actions needed to make assignments ready when scheduled.
Work that is allowed into the lookahead period is evaluated each week before being
permitted to advance further. As stated in Step 3, the goal is to maintain a buffer of roughly two
weeks’ worth of sound assignments, or to adjust the labor force to actual work flow when that
buffer cannot be maintained. A two week buffer has proven to be a practical goal, although some
variation in size can occur without impacting productivity. There may well be cases in which PPC
is consistently high that require less than two weeks’ worth of work for a production unit.
However, to properly size the buffer, it is necessary to first determine the actual productive
capacity of the unit and the extent of variation in productivity over time.
All involved agreed that the next stage in the development process should concentrate
on improving lookahead planning. To help achieve that improvement, some type of performance
measurement is needed. How well lookahead planning was done on the mechanical contractor
pilot project has thus far been assessed in the following ways:
Suppose we are interested in the week of 1/6/97. Consider the work plan for that week to
be the plan for Week One, the lookahead schedule issued on 12/30/96, in which 1/6/97 is the
first of the lookahead weeks, to be the plan for Week Two, and the lookahead schedule issued
12/23/96, in which 1/6/97 is the second of the lookahead weeks, to be the plan for Week Three.
PPC measures the percentage of Week One assignments completed, AA-Wk1 measures the
percentage of those assignments that were anticipated one week ahead, and AA-Wk2 measures
the percentage of those assignments that were anticipated two weeks ahead. As shown in Figure
12, in the six week period from 12/23/96 through 1/27/97, superintendents identified one week
ahead (AA-Wk1) only 52% of the assignments that appeared on weekly work plans in the week
following. In the same period, AA-Wk2 was only 40%; i.e., only 40% of the assignments that
appeared on weekly work plans for 1/6/97 were scheduled for that date on 12/23/97.
9
© 1997 by Glenn Ballard. All Rights Reserved.
Technical Report No. 97-3, Construction Engineering and Management Program,
Civil and Environmental Engineering Department, University of California, Berkeley, CA 94720
Proceedings 5th Annual Conference of the International Group for Lean Construction,
Griffith University, Gold Coast, Australia. July, 1997.
60%
50%
40%
30%
20%
10%
0%
AA-Wk1 AA-Wk2
0.6
0.5
0.4
0.3
0.2
0.1
0
AMR-Wk1 AMR-Wk2
A last method of analysis, time/time charts3, has so far been used sparingly. Scott was a
frontline supervisor working for superintendent Larry on the pilot project. One of Scott’s
assignments on his weekly work plan for 1/20/97 was “Finish AHU 274.1.” As shown in Table 1,
that assignment first appeared on the lookahead schedule Larry issued 12/30/96, and was
scheduled for the week of 1/6. The assignment appeared on Larry’s next lookahead schedule
3
The idea for time/time charts came to the author from Mike Vorster at Virginia Tech by way of
Greg Howell, who also had a hand in the matter.
10
© 1997 by Glenn Ballard. All Rights Reserved.
Technical Report No. 97-3, Construction Engineering and Management Program,
Civil and Environmental Engineering Department, University of California, Berkeley, CA 94720
Proceedings 5th Annual Conference of the International Group for Lean Construction,
Griffith University, Gold Coast, Australia. July, 1997.
(issued 1/6/97) and was scheduled for 1/13. On Larry’s lookahead issued 1/13/97, the
assignment was scheduled for 1/20, and in fact was done on 1/20 according to Scott’s statused
weekly work plan for that date. The time/time charts shown in the tables below have issue dates
vertically and scheduled dates horizontally.
Table 2 shows the changes over time in the schedule for the assignment “Chemical
Feed System”, which the superintendent might have realized was far too big for a single
assignment. In fact, it was eventually divided into multiple parts and completed over a number of
months.
Some details about materials-related problems can be revealed by time/time charts such
as the one shown in Table 3 for the assignment “Raise 200 degree HW on 1st flr West Wing and
reconnect ends.” In the lookahead schedule issued 12/30, the superintendent noted “release for
delivery 1/9.” However, in the next lookahead schedule, he noted “release material for 1-16-97.”
Obviously, the materials could not be acquired when he first wanted them. In the lookahead
issued 1/13, he notes “materials on site” and scheduled the work to be done the week of 1/20. In
fact, the work did not begin until 1/27, and was not entirely completed until the following week for
lack of a control valve and a globe valve, according to the weekly work plan.
11
© 1997 by Glenn Ballard. All Rights Reserved.
Technical Report No. 97-3, Construction Engineering and Management Program,
Civil and Environmental Engineering Department, University of California, Berkeley, CA 94720
Proceedings 5th Annual Conference of the International Group for Lean Construction,
Griffith University, Gold Coast, Australia. July, 1997.
Some assignments were completed precisely on schedule. Table 4 shows one example,
for the assignment “Tie-in HEX-2 on 1st Flr Admin Bldg.” The assignment was not anticipated
until two weeks prior to scheduled start, but its scheduled start remained the same the next
week, and the assignment was completed as scheduled the following week.
1/27
1/13 x
1/20 x
1/27 Completed
Table 4: Time/Time Chart for “Tie-in HEX-2 on 1st Flr Admin Bldg”
All means of evaluating lookahead planning suggest that it needs major improvement.
Far from providing a flow of ready work for selection in weekly work plans, there appears to have
been a disjunct between commitment planning and lookahead planning. This appearance is
supported by examination of the weekly work plans produced by Richard, the front line
supervisor with the highest PPC. Richard divides work into different assignments on his weekly
work plans than did his superintendent on the lookahead schedules that are supposedly his
source of information. Richard often refuses to schedule work when his superintendent obviously
expects him to do so. Richard’s average PPC was approximately 90%, while the next closest
was below 70%. His superintendent has testified that the productivity of Richard’s crew is
consistently equal to or better than that of other crews, indicating that he is not purchasing higher
PPC by selecting less work. In fact, Richard came the closest to following the rule “Only make
workable assignments.” Others made assumptions or placed bets about the soundness of the
assignments listed on their weekly work plans.
Why bother to improve lookahead planning? Why not make all front line supervisors
operate like Richard? For the most part, Richard can protect his crew from failures in work flow
management by refusing to make poor quality assignments, but he cannot overcome the
negative impact on the project of failing to do critical tasks when they need to be done in order to
reduce overall project duration. In addition, he cannot create ready work if the missing
ingredients are either outside his power to acquire or have lead times greater than his personal
‘lookahead’. If sequence were irrelevant and if labor capacity could always be matched to work
load, simply shielding production units in the weekly work planning process would be sufficient.
However, sequence is relevant and productivity can suffer from insufficient load, so lookahead
planning is necessary. How can lookahead planning be improved?
Anticipating assignments would be helped if activities on the master schedule could be
easily divided into operations, which in turn could be divided into assignments by location or
quantity. Master schedule activities tend to be expressed in terms of facility components, often
12
© 1997 by Glenn Ballard. All Rights Reserved.
Technical Report No. 97-3, Construction Engineering and Management Program,
Civil and Environmental Engineering Department, University of California, Berkeley, CA 94720
Proceedings 5th Annual Conference of the International Group for Lean Construction,
Griffith University, Gold Coast, Australia. July, 1997.
divided by location, e.g., “Prefab MP Ductwork-Admin 1flA.” In order to identify the various tasks
that must be performed in order to construct that component in that location, it is necessary to
design at least a rudimentary work method.
As regards making assignments ready, to the extent that a bill of materials and design
documents can be associated with each scheduled activity, the ultimate assignment can be
made sound as regards materials and design. However, it is not so easy to identify the temporary
structures and materials needed or to determine the prerequisite work that must be completed
beforehand. These require thinking through the scheduled activity as an operation or process,
i.e., deciding how the work will be done, in what circumstances, with what material staging areas,
with what adjacent trades, etc. The work of process modelers might be helpful in this regard,
specifically those who correlate standard processes with facility components and build up master
schedule activities from those processes (Odeh 1992). In any case, it appears that a key to
improved lookahead planning is methods planning (the subject of later papers in this research
program).
The objectives posed for lookahead planning were to shape work flow, match load and
capacity, maintain a backlog of sound assignments, and identify operations. Pilot project
lookahead planning did not adequately accomplish any of these objectives. Procedures were not
sufficiently developed and superintendents were not adequately trained. Superintendents did not
have the necessary information and neither did their “suppliers” (detailers, fabrication shops,
vendors, trades doing prerequisite work, project engineers, etc) have the information needed to
coordinate fabrication and deliveries with assignments that were otherwise sound.
Superintendents and their “suppliers” did not share the same model of production control and
consequently were not fully cooperative. The amount of time and energy available for planning
may also have been a factor. Planning and control are traditionally understaffed on construction
projects, and no changes in staffing were made to support the new production control system.
References
Ballard, G. and Howell, G. (1994). “Implementing Lean Construction: Stabilizing Work Flow.”
Proceedings of the 2nd Annual Meeting of the International Group for Lean Construction,
Santiago, Chile. (Available in Lean Construction, A.A. Balkema Publishers, Rotterdam,
Netherlands, 1997.)
Ballard, G. and Howell, G. (1995). “Toward Construction JIT.” Proceedings of the 1995 ARCOM
Conference, Association of Researchers in Construction Management, Sheffield, England.
13
© 1997 by Glenn Ballard. All Rights Reserved.
Technical Report No. 97-3, Construction Engineering and Management Program,
Civil and Environmental Engineering Department, University of California, Berkeley, CA 94720
Proceedings 5th Annual Conference of the International Group for Lean Construction,
Griffith University, Gold Coast, Australia. July, 1997.
14