4.2 Resource Leveling vs. Resource Smoothing

4.2 Resource Leveling vs. Resource Smoothing

AACE Professional Context: One of the most frequently tested distinctions on the AACE Planning and Scheduling Professional (PSP) examination is the technical difference between Resource Leveling and Resource Smoothing. While lay practitioners often use these terms interchangeably, in rigorous cost engineering and project controls they represent fundamentally opposing optimization constraints. Leveling prioritizes resource availability limits over completion time, whereas smoothing prioritizes completion time over resource demand. Schedulers must understand how these methods interact with project float, how leveling alters critical paths into Resource Critical Paths (RCP), and the operational hazards of activity splitting.

---

The Core Distinction: Leveling vs. Smoothing

The fundamental divide between Resource Leveling and Resource Smoothing centers on which project parameter is treated as a hard constraint (inviolable) and which is treated as a soft constraint (flexible).

                    [UNCONSTRAINED SCHEDULE (OVER-ALLOCATION)]
                                  Peak Demand
                                 ┌───────────┐
                                 │ Over-Cap  │
                    ─────────────┼───────────┼───────────── [Resource Limit]
                                 │ Base Load │
                    ─────────────┴───────────┴─────────────► Time
                                       │
         ┌─────────────────────────────┴─────────────────────────────┐
         ▼                                                           ▼
  [RESOURCE LEVELING]                                         [RESOURCE SMOOTHING]
- Limit is a HARD ceiling                                   - Project End Date is a HARD wall
- Activities pushed out                                     - Activities shifted ONLY within float
- Project Duration EXTENDS                                  - Project Duration CANNOT extend

            Limit                                                       Limit
  ──────────┬───────┬────────                         ──────────┬───/\───┬──────── (Peak may breach)
            │ Level │                                           │ Smooth│
  ──────────┴───────┴───────► Time                    ──────────┴───────┴────────► Time
                     ▲                                              ▲
            [LATER FINISH DATE]                            [SAME FINISH DATE]

---

Algorithmic Priority Heuristics in Resource Leveling

Resource leveling is a computationally complex problem. In mathematics and computer science, finding the absolute optimal sequence that minimizes project duration under multi-resource constraints is proven to be NP-hard (non-deterministic polynomial-time hard). Exact analytical optimization is practically impossible for real-world project networks containing thousands of activities and dozens of shared resource pools.

Consequently, all major commercial scheduling software engines (e.g., Primavera P6, Microsoft Project, Asta Powerproject) utilize heuristic forward-pass simulation rules to resolve resource contentions. When multiple activities compete for the same scarce resource on a given working day, the leveling engine applies a defined Priority Rule to determine which activity receives the resource and which activity must wait.

The Standard Priority Heuristics

1. Minimum Total Float (Smallest Slack / Early Late Finish):
   • Mechanism: The activity with the smallest positive total float (or earliest Late Start / Late Finish date) is given first priority to the resource. Activities with larger float are delayed.
   • Engineering Rationale: This is the most widely utilized and mathematically defensible heuristic in project controls. Prioritizing activities with the least float minimizes the likelihood and magnitude of overall project completion delays.
2. Early Start Date (First-Come, First-Served):
   • Mechanism: The activity that was logically eligible to start earliest in the forward pass receives priority.
   • Engineering Rationale: Preserves the natural chronological build sequence of the project, though it may occasionally delay near-critical activities in favor of non-critical tasks that started earlier.
3. Shortest Activity Duration:
   • Mechanism: The competing activity with the shortest duration is scheduled first.
   • Engineering Rationale: Based on queuing theory (Shortest Processing Time First). Rapidly clearing short tasks frees up the resource pool quickly for larger, subsequent activities.
4. Greatest Resource Demand / Intensity:
   • Mechanism: The activity that requires the highest number of resource units per day is given priority.
   • Engineering Rationale: Prevents resource-heavy "monster" activities from being repeatedly postponed to the end of the project where they cannot be accommodated.
5. Activity Priority Codes (User-Defined Ranking):
   • Mechanism: Planners manually assign numerical priorities (e.g., Priority 1 = Top, Priority 10 = Low) based on external operational risks, permitting windows, or commercial milestones.
   • Engineering Rationale: Gives the project controls team direct governance over software decision-making for strategic project components.

---

The Resource-Constrained Critical Path (Resource Critical Path - RCP)

In classical CPM mechanics, the critical path is strictly defined as the longest continuous chain of logically linked activities from project start to finish, characterized by zero (or minimum) total float. This definition assumes that precedence arrows represent physical construction sequencing (e.g., foundations must cure before structural steel is erected).

When a schedule is leveled under resource constraints, this classical definition breaks down, giving rise to the Resource Critical Path (RCP).

Phantom Dependencies (Resource Transfer Logic)

Consider two activities on completely separate physical branches of a project network:

• Activity X (Turbine Foundation Excavation): Located in the power block.
• Activity Y (Retention Basin Excavation): Located 1,000 meters away in the offsites area.

Physically, there is zero engineering logic connecting Activity X and Activity Y; they can proceed concurrently. However, if the project has only one large excavator, the leveling engine will force Activity Y to wait until Activity X finishes.

This introduces a Resource Transfer Dependency (often termed a phantom dependency or soft logic tie):

$\text{Activity X (Power Block)} \xrightarrow{\text{Resource Transfer: Excavator}} \text{Activity Y (Retention Basin)}$

[PHYSICAL CPM NETWORK (Unleveled)]        [RESOURCE-LEVELING RESULT (Leveled)]
Path 1: Act X (Excavate Power Block)      Path 1: Act X (Excavate Power Block)
                                                    │
                                                    ▼ [Resource Dependency: Excavator]
Path 2: Act Y (Excavate Retention Basin)  Path 2: Act Y (Excavate Retention Basin)
(Independent, parallel paths)             (Paths serialized by shared resource!)

Structural Properties of the Resource Critical Path

1. Mixed Logic Chains: The RCP is composed of a hybrid sequence of physical engineering dependencies (FS, SS, FF) and resource-driven transfer dependencies.
2. Invalidation of Traditional Float: In an unconstrained network, total float is calculated as $TF = LS - ES$. In a leveled schedule, this mathematical value is misleading. An activity may appear to have 15 days of traditional float based on physical successors, but it has zero real float because delaying it would starve a downstream task of its required crew, causing project completion delay. The true float is Resource Total Float (RTF).
3. Forensic significance: If credible resource constraints drove sequence, a delay analysis should test those resource effects rather than rely only on an unconstrained logic path. RP 29R-03 does not impose one universal resource-critical-path method; the contract, records, and chosen analysis protocol control. An owner delay to an activity that holds a driving resource can delay the entire project, even if that activity was technically "non-critical" in the original unconstrained logic network.

---

Activity Splitting Mechanics and Operational Hazards

During resource leveling, schedulers can enable a software parameter known as Activity Splitting (or Activity Preemption).

How Activity Splitting Operates

• When an in-progress activity encounters a resource shortage—or when a higher-priority task requires the same resource—the leveling engine pauses (splits) the lower-priority activity.
• The activity's remaining duration is held in suspension while its assigned resources are temporarily reassigned elsewhere.
• Once the higher-priority task completes and resources become available again, the split activity resumes execution.

UNSPLIT:  [===== 8 Continuous Days =====]

SPLIT:    [== 3 Days ==] ... [PAUSED: 4 Days] ... [===== 5 Days Remaining =====]
                         ▲                         ▲
                   Demobilize                 Re-mobilize
                   Resources                  Resources

When Activity Splitting is Legitimate

Splitting is acceptable for non-continuous, easily interrupted tasks:

• Software development, engineering drawing production, or administrative submittal preparation.
• Off-site modular fabrication where pieces can sit in staging yards without deterioration.
• Broad earthworks grading where equipment can be redirected between borrow pits without technical harm.

Severe Operational Risks and Hazards of Splitting

On industrial, infrastructure, and commercial construction projects, indiscriminate activity splitting creates severe operational disruption:

1. Setup and Demobilization Penalties: Re-mobilizing equipment is costly and time-consuming. If an activity requiring a 250-ton crane is split, the contractor incurs crane teardown, relocation, and re-rigging costs, along with re-inspection of rigging tackle.
2. Physical Quality Degradation: Many construction operations cannot tolerate work stoppages without destroying structural integrity:
   • Concrete Pours: Splitting a continuous structural foundation pour results in an unapproved cold joint, compromising the shear strength and watertightness of the structure.
   • Welding Operations: Pausing heavy-wall alloy pipe welding requires cooling, re-heating (preheat/PWHT cycles), and extensive non-destructive re-examination (radiography/ultrasonic testing).
   • Trench Excavation: Leaving an open trench split for two weeks risks sidewall collapse, water accumulation, and safety violations under OSHA trenching standards.
3. Productivity Degradation (Learning Curve Decay): When a crew is pulled off an active installation, they lose their daily rhythm, trade momentum, and tooling setup familiarity. Upon resumption, the crew suffers a learning curve reset, burning more hours to complete the remaining scope.
4. Subcontractor Delay and Disruption Claims: In fixed-price or unit-rate subcontracts, contractors base their bids on continuous, uninterrupted access to the workface. If a prime contractor repeatedly splits a subcontractor's activities to reassign shared equipment, the subcontractor will file commercial claims for idle labor, equipment standby, loss of productivity, and extended overhead.

---

Worked Example: Step-by-Step Leveling with Minimum Total Float

Consider an industrial mechanical contractor executing piping modifications during plant operations. Two concurrent activities compete for the same scarce resource: Certified High-Pressure Rigging Crews.

Project Parameters

• Available Resource Capacity: Exactly 1 Rigging Crew available per day.
• Timeline: Both activities are logically ready to execute starting on Day 10.
• Activity 101 (Set Heat Exchanger E-101):
  • Duration = 3 working days
  • Resource Requirement = 1 Rigging Crew/day
  • Early Dates: ES = Day 10, EF = Day 13
  • Late Dates: LS = Day 10, LF = Day 13
  • Total Float ($TF$) = $LS - ES = 10 - 10 =$ 0 days (Critical Path)
• Activity 102 (Erect Flare Knockout Drum Platform):
  • Duration = 4 working days
  • Resource Requirement = 1 Rigging Crew/day
  • Early Dates: ES = Day 10, EF = Day 14
  • Late Dates: LS = Day 16, LF = Day 20
  • Total Float ($TF$) = $LS - ES = 16 - 10 =$ 6 days

Step 1: Detect Resource Contention

On Days 10, 11, and 12, both Activity 101 and Activity 102 demand 1 Rigging Crew. $\text{Total Demand} = 1 + 1 = 2\text{ crews} > 1\text{ crew available (OVER-ALLOCATION)}$

Step 2: Apply Priority Heuristic (Minimum Total Float)

The leveling engine evaluates the competing tasks on Day 10 using the Minimum Total Float rule:

• Activity 101 has $TF = 0$ days.
• Activity 102 has $TF = 6$ days.
• Decision: Activity 101 has smaller float ($0 < 6$). Activity 101 receives the Rigging Crew on Day 10. Activity 102 must be delayed.

Step 3: Forward Pass Simulation

• Activity 101 Executes: Operates from Day 10 through Day 13. The crew is fully utilized.
• Activity 102 Waits: Delayed from Day 10 until Day 13 when Activity 101 releases the rigging crew.
• Activity 102 Resumes: Starts on Day 13 and finishes on Day 17 ($13 + 4 = 17$).

Step 4: Evaluate Impact on Float and Project Duration

1. Delay Incurred by Activity 102: Shifted from Day 10 to Day 13 (a 3-day delay).
2. Float Consumption: Activity 102 had 6 days of total float. Consuming 3 days of float leaves: $\text{Remaining Total Float} = 6 - 3 = 3\text{ days}$
3. Project Completion Date Impact: Because Activity 102 still has 3 days of positive float remaining upon completion at Day 17 (its late finish was Day 20), the project completion date did NOT slip by a single day.
4. Classification: This successful resolution represents Resource Smoothing within available float.

Contrast Scenario: Resource Leveling with Schedule Extension

What if Activity 102 also had $TF = 0$ days?

• Delaying Activity 102 by 3 days would immediately force its Late Finish from Day 14 to Day 17, extending project completion by 3 working days.
• This secondary scenario represents Resource Leveling (Resource-Constrained Scheduling).