2.3 Forward and Backward Pass Mathematical Mechanics

2.3 Forward and Backward Pass Mathematical Mechanics

The Critical Path Method executes two sequential mathematical algorithms across the network model: the Forward Pass and the Backward Pass. Together, these calculations determine the earliest and latest possible dates each activity can occur without extending the overall project duration.

---

The CPM Activity Block Layout

In CPM precedence diagramming, activity data is traditionally presented in a standardized six-compartment block (or node):

┌──────────────┬──────────────┬──────────────┐
│ Early Start  │   Duration   │ Early Finish │
│     (ES)     │     (D)      │     (EF)     │
├──────────────┴──────────────┴──────────────┤
│           Activity ID & Name               │
├──────────────┬──────────────┬──────────────┤
│  Late Start  │ Total Float  │ Late Finish  │
│     (LS)     │     (TF)     │     (LF)     │
└──────────────┴──────────────┴──────────────┘

---

Zero-Based vs. One-Based Date Calculation Conventions

A primary source of calculation error on scheduling exams and in field practice is confusion between the zero-based convention (end-of-day / continuous timeline) and the one-based convention (calendar workday numbers).

Why Both Exist

• Zero-Based Convention: A useful continuous-time convention for hand calculations: a 5-day activity starting at time 0 finishes at time 5.0 ($0 + 5 = 5$), when its zero-lag FS successor can begin. Commercial tools use product-specific calendar, time-of-day, and display settings, so confirm those settings before reconciling dates.
• One-Based Convention: Used when presenting calendar dates to human field teams. If an activity starts on the morning of Monday (Day 1) with a 5-day duration, it finishes on the evening of Friday (Day 5), calculated as $1 + 5 - 1 = 5$. The successor starts on the morning of the next working day, Day 6 ($5 + 1 = 6$).

---

The Forward Pass Algorithm

The forward pass determines the Early Start (ES) and Early Finish (EF) dates. It calculates the earliest possible project completion date.

Step-by-Step Forward Pass Rules:

1. Initialize Project Start: For initial activities with no predecessors, set $ES = 0$ (zero-based).
2. Calculate Early Finish: $EF = ES + \text{Duration}$.
3. Traverse Network: Move sequentially from left to right across dependency links.
4. Evaluate Relationship Constraints:
   • Finish-to-Start (FS): $ES_{\text{succ}} \ge EF_{\text{pred}} + \text{Lag}_{FS}$
   • Start-to-Start (SS): $ES_{\text{succ}} \ge ES_{\text{pred}} + \text{Lag}_{SS}$
   • Finish-to-Finish (FF): $EF_{\text{succ}} \ge EF_{\text{pred}} + \text{Lag}_{FF}$
   • If constrained by an FF relationship, the resulting Early Start is: $ES_{\text{succ}} = EF_{\text{succ}} - \text{Duration}$.
5. Resolve Merge Nodes (Convergence): When an activity has multiple incoming predecessors, its Early Start is governed by the maximum (latest) calculated constraint:

$ES_j = \max_{i} \left( \text{Constraint}_i \right)$

The predecessor that creates the maximum value is the driving predecessor.

---

The Backward Pass Algorithm

The backward pass determines the Late Finish (LF) and Late Start (LS) dates. It calculates the latest dates activities can occur without delaying the overall project completion date.

Step-by-Step Backward Pass Rules:

1. Initialize Project Completion: For terminal activities with no successors, set Late Finish equal to Early Finish ($LF = EF$), unless an external contract completion deadline is imposed.
2. Calculate Late Start: $LS = LF - \text{Duration}$.
3. Traverse Network: Move sequentially from right to left across dependency links.
4. Evaluate Relationship Constraints:
   • Finish-to-Start (FS): $LF_{\text{pred}} \le LS_{\text{succ}} - \text{Lag}_{FS}$
   • Start-to-Start (SS): $LS_{\text{pred}} \le LS_{\text{succ}} - \text{Lag}_{SS}$
   • If constrained by an SS relationship, the resulting Late Finish is: $LF_{\text{pred}} = LS_{\text{pred}} + \text{Duration}$.
   • Finish-to-Finish (FF): $LF_{\text{pred}} \le LF_{\text{succ}} - \text{Lag}_{FF}$
5. Resolve Burst Nodes (Divergence): When an activity has multiple outgoing successors, its Late Finish is governed by the minimum (earliest) calculated constraint:

$LF_i = \min_{j} \left( \text{Constraint}_j \right)$

The successor that creates the minimum value is the driving successor.

---

Comprehensive Worked Numerical Example: 5-Activity Network

Let us calculate a 5-activity network containing standard Finish-to-Start links, a Start-to-Start lag, and a Finish-to-Finish lag, using the zero-based convention.

Activity Definitions & Logic:

• Activity A: "Mobilization & Site Clearing", Duration = 4 days. Predecessors: None.
• Activity B: "Bulk Excavation", Duration = 6 days. Predecessor: Activity A (FS = 0).
• Activity C: "Underground Utility Trenching", Duration = 5 days. Predecessor: Activity A (SS with lag = +2 days).
• Activity D: "Foundation Concrete Work", Duration = 7 days. Predecessors:
  • Predecessor 1: Activity B (FS = 0)
  • Predecessor 2: Activity C (FF with lag = +3 days)
• Activity E: "Backfill & Grading", Duration = 3 days. Predecessor: Activity D (FS = 0).

                    ┌───► [Activity B (D=6)] ──(FS=0)───┐
                    │                                   ▼
[Activity A (D=4)] ─┤                              [Activity D (D=7)] ──(FS=0)──► [Activity E (D=3)]
                    │                                   ▲
                    └───► [Activity C (D=5)] ──(FF=+3)──┘
                        (SS = +2)

Forward Pass Step-by-Step:

1. Activity A (Duration = 4):

   • Has no predecessors $\rightarrow ES_A = 0$.
   • $EF_A = ES_A + D_A = 0 + 4 = \mathbf{4}$.

2. Activity B (Duration = 6):

   • Predecessor A via FS = 0 $\rightarrow ES_B = EF_A = \mathbf{4}$.
   • $EF_B = ES_B + D_B = 4 + 6 = \mathbf{10}$.

3. Activity C (Duration = 5):

   • Predecessor A via SS with lag +2 $\rightarrow ES_C = ES_A + 2 = 0 + 2 = \mathbf{2}$.
   • $EF_C = ES_C + D_C = 2 + 5 = \mathbf{7}$.

4. Activity D (Duration = 7) [Merge Node]:

   • From Predecessor B (FS = 0): Requires $ES_D \ge EF_B = 10$.
   • From Predecessor C (FF = +3): Requires $EF_D \ge EF_C + 3 = 7 + 3 = 10$. This implies $ES_D \ge EF_D - D_D = 10 - 7 = 3$.
   • Evaluate maximum constraint for Early Start: $ES_D = \max(10, 3) = \mathbf{10}$.
   • Calculate Early Finish: $EF_D = ES_D + D_D = 10 + 7 = \mathbf{17}$.
   • (Driving Predecessor is Activity B).

5. Activity E (Duration = 3):

   • Predecessor D via FS = 0 $\rightarrow ES_E = EF_D = \mathbf{17}$.
   • $EF_E = ES_E + D_E = 17 + 3 = \mathbf{20}$.
   • Project Early Completion Date = Day 20.

Backward Pass Step-by-Step:

1. Activity E (Duration = 3):

   • Set $LF_E = EF_E = \mathbf{20}$.
   • $LS_E = LF_E - D_E = 20 - 3 = \mathbf{17}$.

2. Activity D (Duration = 7):

   • Successor E via FS = 0 $\rightarrow LF_D = LS_E = \mathbf{17}$.
   • $LS_D = LF_D - D_D = 17 - 7 = \mathbf{10}$.

3. Activity B (Duration = 6):

   • Successor D via FS = 0 $\rightarrow LF_B = LS_D = \mathbf{10}$.
   • $LS_B = LF_B - D_B = 10 - 6 = \mathbf{4}$.

4. Activity C (Duration = 5):

   • Successor D via FF with lag +3 $\rightarrow$ Requires $LF_C \le LF_D - 3 = 17 - 3 = 14$.
   • $LS_C = LF_C - D_C = 14 - 5 = \mathbf{9}$.

5. Activity A (Duration = 4) [Burst Node]:

   • To Successor B via FS = 0: Requires $LF_A \le LS_B = 4$.
   • To Successor C via SS with lag +2: Requires $LS_A \le LS_C - 2 = 9 - 2 = 7$, which implies $LF_A \le LS_A + D_A = 7 + 4 = 11$.
   • Evaluate minimum constraint for Late Finish: $LF_A = \min(4, 11) = \mathbf{4}$.
   • $LS_A = LF_A - D_A = 4 - 4 = \mathbf{0}$.
   • (Driving Successor is Activity B).

Network Summary Table

Critical Path: $A \rightarrow B \rightarrow D \rightarrow E$ (Total Duration = 20 days). Non-Critical Activity: Activity C has Total Float = 7 days ($9 - 2 = 7$). Its FF requirement ($EF_C + 3 = 10$) was non-driving because B pushed Activity D to Day 10.