Work structuring, location-based management and buffers

Takt planning and the Last Planner System both depend on decisions made before any schedule is drawn: how the work is divided into pieces, how the building is divided into places, and where slack is held to absorb the variability that no plan can remove. Lean construction calls the first set of decisions work structuring. Location-based management, developed in Finland by Olli Seppänen and Russell Kenley among others, provides a way of planning and controlling work by location that is especially useful when trades do not all move at one rhythm. And buffers, the deliberate slack in a production system, determine how well a plan survives contact with reality. This lesson covers all three.

Work structuring

Work structuring is the design of the production system itself: deciding how to break the project into pieces of work, who will do each piece, in what order and in what batch sizes, and where the hand-offs between trades will be. In traditional practice these decisions are made implicitly, by the way the estimate is organized into trade packages and the schedule into activities. Lean practice makes them explicit and makes them together with the trades. Good work structuring aims for pieces of work that one crew can complete without interruption, hand-offs that are few and clear, and batch sizes small enough that work flows.

One useful idea is to separate work by the kind of flow it follows. Work in repeated spaces suits takt. Work in unique spaces suits a conventional critical path schedule. Long-lead procurement and fabrication follow a supply chain flow that must be pulled to match installation. Recognizing these flows early lets each be planned with the right tool and lets the interfaces between them be managed deliberately.

The location breakdown structure

Location-based planning starts with a location breakdown structure (LBS): a hierarchy of places, such as building, floor and zone, analogous to the work breakdown structure of a CPM schedule (CON-210). Each trade's work is then quantified by location, so that the planner knows how many hours of framing, rough-in or tile each zone needs. Because quantities often differ between locations, for example a ground floor with retail space and an upper floor with suites, the planner can see where each trade will take longer and adjust crews or sequences accordingly.

Flowlines and location-based control

In the location-based management system (LBMS), each trade is drawn as a flowline: a line through the locations over time, like a line of balance chart. The planner aims for three things: continuous flow, so each crew moves from location to location without stopping; parallel flowlines, so trades work at matching rates; and no crossing lines, which would mean two trades in the same place at once. During construction, actual progress by location is recorded and plotted against the plan. Because each trade's actual production rate is known, the system can forecast where it will be in the coming weeks and raise an alarm if two flowlines are forecast to collide, well before the collision happens.

# Location-based forecast: will drywall catch rough-in? (floors 2..8, illustrative)
# Rough-in actual: started floor 2 at week 0; has finished 3 floors by week 4 -> actual rate 0.75 floors/wk
# Drywall plan: start floor 2 at week 3, planned rate 1.0 floor/wk, needs 1 week buffer behind rough-in
rough_start, rough_done_floors, now = 0.0, 3, 4.0
rough_rate = rough_done_floors / (now - rough_start)
dry_start, dry_rate, buffer = 3.0, 1.0, 1.0
print(f"Rough-in actual rate: {rough_rate:.2f} floors per week")
for i, floor in enumerate(range(2, 9)):
    rough_finish = rough_start + (i + 1) / rough_rate          # forecast finish of this floor
    dry_start_floor = dry_start + i / dry_rate                 # planned drywall start on this floor
    gap = dry_start_floor - rough_finish
    flag = "  <-- conflict" if gap < buffer else ""
    print(f"floor {floor}: rough-in done wk {rough_finish:5.2f}  drywall planned start wk {dry_start_floor:5.2f}  gap {gap:5.2f}{flag}")

Rough-in actual rate: 0.75 floors per week
floor 2: rough-in done wk  1.33  drywall planned start wk  3.00  gap  1.67
floor 3: rough-in done wk  2.67  drywall planned start wk  4.00  gap  1.33
floor 4: rough-in done wk  4.00  drywall planned start wk  5.00  gap  1.00
floor 5: rough-in done wk  5.33  drywall planned start wk  6.00  gap  0.67  <-- conflict
floor 6: rough-in done wk  6.67  drywall planned start wk  7.00  gap  0.33  <-- conflict
floor 7: rough-in done wk  8.00  drywall planned start wk  8.00  gap  0.00  <-- conflict
floor 8: rough-in done wk  9.33  drywall planned start wk  9.00  gap -0.33  <-- conflict

The forecast shows that, at rough-in's actual rate, drywall will have exactly its one-week buffer on floor 4 and will run short of it from floor 5 onward, until by floor 8 it would be trying to start before rough-in has finished. The team has several weeks' warning and can choose a response: add a rough-in crew, slow the drywall start so its crew works continuously behind rough-in, or resequence. Location-based control turns a vague sense that rough-in is behind into a specific forecast of where and when the problem will hit.

Buffers

A buffer is slack deliberately placed in a production system to absorb variability. Lesson 3 showed that variability among dependent trades reduces throughput unless something absorbs it. That something can be extra time, extra capacity, extra inventory of ready work or extra space. Each kind protects the flow in a different way and has a different cost.

The principle is to reduce variability first and then place buffers where the remaining variability is highest and does the most damage, sized to the level of protection needed. Traditional projects have plenty of buffers, but they are hidden: padding in each trade's duration, stockpiles of material, floors left idle. Lean planning makes buffers visible, pools them and manages their consumption.

Sizing a time buffer

How much time buffer is needed between two trades depends on how variable the upstream trade's completion is and on how much protection the downstream trade needs. A simulation can show the trade-off: for each buffer size, how often does the downstream trade find its zone not ready?

# How often is the next zone not ready for drywall? Effect of a time buffer (seeded simulation)
import random
rng = random.Random(410)
def late_prob(buffer_days, n=20000):
    late = 0
    for _ in range(n):
        # upstream wagon planned at 2.0 days per zone; actual: usually on time, sometimes late
        actual = 2.0 + max(-0.5, rng.gauss(0.2, 0.6)) + (rng.random() < 0.08) * rng.uniform(1, 3)
        if actual > 2.0 + buffer_days:
            late += 1
    return late / n
for b in (0, 0.5, 1.0, 2.0, 3.0):
    print(f"buffer {b:>3} days: downstream zone not ready {late_prob(b):6.1%} of the time")

buffer   0 days: downstream zone not ready  65.4% of the time
buffer 0.5 days: downstream zone not ready  37.0% of the time
buffer 1.0 days: downstream zone not ready  15.3% of the time
buffer 2.0 days: downstream zone not ready   4.8% of the time
buffer 3.0 days: downstream zone not ready   1.3% of the time

With no buffer, the board crew would find its zone not ready about two times in three, because the upstream wagon usually runs a little late. A half-day buffer cuts that to a little over one in three, a one-day buffer to about one in seven and a two-day buffer to about one in twenty, but the last few percent come from occasional large delays that a time buffer of practical size cannot cover. Those are better attacked at the source, through make-ready planning and constraint removal, than by ever-larger buffers. Each day of buffer between every pair of wagons would add many days to the overall schedule, so buffers belong where they are most needed, typically before trades with long set-up, after inspections and where hand-offs have historically been unreliable.

Decoupling and hand-offs

A buffer decouples two steps: it lets the upstream trade's variability be absorbed before it reaches the downstream trade. Good places to decouple are where the upstream work is highly variable (for example after an inspection by a third party), where the downstream work has a long set-up (for example a tile setter who must prepare substrates and materials) and between groups of trades from different companies with different rhythms. Between closely coordinated trades in a stable takt, buffers can be small or absent. Formal hand-offs, in which the upstream trade releases a zone with a short checklist and the downstream trade accepts it, reduce the variability that buffers must absorb by catching incomplete work before the next trade starts.

Location-based planning and the critical path

Location-based planning does not abandon network logic. Dependencies still exist: rough-in before board, board before tape. What changes is that dependencies are expressed between locations as well as between activities, and that continuity of crews is treated as a goal in itself. Some scheduling software supports both views, so that the critical path is calculated with location-based logic and crews are scheduled to work continuously where possible. The contract schedule required by the owner can still be produced in CPM form, while the field is managed by location.