11.2 Assembly Line Balancing, Takt Time, Cycle Time, and Throughput

Assembly lines are product-oriented production flow systems where discrete work units move sequentially along a configured material handling transfer system past assigned workstations. Balancing an assembly line involves partitioning the total work content into discrete subsets assigned to workstations such that technological precedence constraints are satisfied and station idle times are minimized.

---

1. Flow Line Principles: Demand Pacing vs. Capacity

To configure an assembly line, industrial engineers must distinguish between the pace required by the market and the operational capability of the manufacturing line.

Line Balancing Parameter Relationships:
 ├── Customer Demand (D) & Available Time (T_a) ──> Takt Time = T_a / D (Pacing heartbeat)
 ├── Bottleneck Station Duration (S_max) ─────────> Line Cycle Time C = S_max (Production interval)
 ├── Production Pacing Condition ────────────────> C <= Takt (Prevents starving customer demand)
 └── Line Capacity / Throughput ──────────────────> TH = 1 / C (Units produced per unit time)

Takt Time ($Takt$)

Originating from the German word for "musical meter" or "beat," Takt time represents the rate at which finished goods must emerge from the line to exactly satisfy customer demand without building excess finished-goods inventory:

$Takt = \frac{\text{Available Net Operating Time per Period} (T_a)}{\text{Customer Demand per Period} (D)}$

Critical Convention: Available operating time ($T_a$) must account only for planned production operating duration. Unpaid lunch breaks, scheduled preventive maintenance, and planned plant shutdowns must be subtracted from the gross shift length. If paid rest breaks occur while the line is halted, that break time must also be deducted to find the net productive time.

Cycle Time ($C$)

Cycle time ($C$ or $T_c$) is the actual elapsed time between consecutive finished units exiting the final workstation of the line. On a synchronous transfer line without inter-station buffer accumulation, the overall line cycle time is governed by the slowest station—the bottleneck station:

$C = \max_{j=1,\dots,N} \{S_j\}$

Where $S_j$ is the total work element processing time assigned to station $j$ ($S_j = \sum_{i \in \text{Station } j} t_i$).

• Stability and Feasibility Requirement: To fulfill customer demand without overtime, the line cycle time cannot exceed Takt time: $C \le Takt$.
• Throughput Rate ($TH$): The continuous output rate of the production line: $TH = \frac{1}{C}$ (units per unit time).

---

2. Precedence Constraints & Theoretical Minimum Workstations

Precedence Diagram and Total Work Content

Every product design requires a specific set of basic work elements (tasks) $i = 1, 2, \dots, n$, each having a known deterministic execution duration $t_i$. The total work content ($T_{\text{wc}}$) is the arithmetic sum of all individual task times:

$T_{\text{wc}} = \sum_{i=1}^n t_i$

A precedence diagram is a directed acyclic graph (DAG) where nodes represent individual work elements and directed arrows represent technological sequencing constraints (e.g., a screw cannot be torqued until the washer and bracket are placed).

Illustrative Precedence Graph:
    [Task 1: 0.8 min] ───> [Task 3: 0.5 min] ──┐
                                              ├───> [Task 5: 1.2 min]
    [Task 2: 1.0 min] ───> [Task 4: 0.9 min] ──┘

Theoretical Minimum Number of Workstations ($N_{\min}$)

Dividing total work content by the desired cycle time establishes the theoretical lower bound on required workstations:

$N_{\min} = \left\lceil \frac{\sum_{i=1}^n t_i}{C} \right\rceil = \left\lceil \frac{T_{\text{wc}}}{C} \right\rceil$

Where $\lceil \cdot \rceil$ denotes the mathematical ceiling function, which mandates rounding up to the next integer regardless of how small the fractional remainder may be. Because fractional workstations (e.g., 4.1 stations) cannot exist physically, at least 5 workstations are required.

---

3. Heuristic Assembly Line Balancing Algorithms

The assembly line balancing problem is an NP-hard combinatorial optimization problem. For industrial-scale applications, heuristic algorithms generate near-optimal station assignments efficiently.

Ranked Positional Weight (RPW / Helgeson-Birnie Method)

The RPW method prioritizes tasks that have large processing times and command long chains of dependent downstream tasks.

• Step 1: Compute Positional Weight ($PW_i$) for each task $i$: The positional weight is the sum of task $i$'s own duration plus the durations of all tasks that technologically follow it in the precedence network: $PW_i = t_i + \sum_{k \in \text{Followers}(i)} t_k$
• Step 2: Rank Tasks: Sort all tasks in strictly descending order of their $PW_i$ values.
• Step 3: Assign Tasks to Stations:
  • Open Station 1 with available time capacity equal to $C$.
  • Scan the ranked list from highest to lowest $PW$. Assign task $i$ to the active station if:
    1. All immediate technological predecessors of task $i$ have already been assigned (either to earlier stations or earlier in the current station).
    2. The task duration $t_i$ does not exceed the remaining unassigned time of the active station ($t_i \le C - S_{\text{active}}$).
  • If no eligible unassigned task fits into the remaining station time, close the current station, open Station $j+1$, and repeat until all $n$ tasks are assigned.

Largest Candidate Rule

A simpler heuristic that sorts all unassigned tasks in descending order of individual task times ($t_i$). At each step, the algorithm selects the largest eligible task that satisfies precedence constraints and fits within remaining station capacity. While computationally simpler, the Largest Candidate Rule can become trapped in local sub-optima if a short task blocks a long downstream critical path.

---

4. Line Balancing Metrics: Efficiency, Delay, and Smoothing

Once an assembly line is configured with $N_{\text{actual}}$ workstations and cycle time $C$, its operational performance is quantified using standardized NCEES efficiency metrics:

---

5. Little's Law and Flow Dynamics in Production Lines

In manufacturing systems, Work-in-Process ($WIP$), Throughput ($TH$), and Manufacturing Lead Time ($MLT$ or Flow Time, $CT$) are fundamentally linked by Little's Law:

$WIP = TH \times CT$

Where:

• $WIP$ = average number of parts residing within the line boundaries (in stations and intermediate buffers)
• $TH$ = steady-state average production output rate ($TH = 1 / C$ when running at bottleneck capacity)
• $CT$ = average total flow time (lead time) an individual workpiece spends entering, traversing, and exiting the line

Production Line WIP Buffer Dynamics:
 [Station 1] ───> [Buffer 1] ───> [Station 2 (Bottleneck)] ───> [Buffer 2] ───> [Station 3]
 (S1 = 1.2 min)    (Accumulates)    (S2 = 2.0 min)               (Starved)       (S3 = 1.4 min)

Bottleneck Blocking and Starvation Dynamics

When station cycle times are unevenly balanced:

• Starvation: A downstream station with processing time $S_{j+1} < S_j$ completes its assigned task and remains idle because the slower upstream feeding station has not released the next part.
• Blocking: An upstream station with processing time $S_{j-1} < S_j$ completes its task but cannot discharge the part because the downstream station's buffer is completely full.
• Workload Smoothing Strategy: To alleviate bottlenecks without adding stations, industrial engineers re-sequence tasks across boundaries, introduce parallel workstations at the bottleneck ($C_{\text{effective}} = S_{\text{bottleneck}} / m$), or invest in automated poka-yoke fixturing to compress high-variance task times.

---

6. Step-by-Step Worked Engineering Calculations

Worked Example 11.2.1: Full Line Balancing via Ranked Positional Weight (RPW)

Problem: A manufacturer of medical infusion pumps operates an assembly line for an 8-hour shift. The shift includes two 15-minute paid rest breaks during which the line is stopped. Target customer demand is $D = 300\text{ units/shift}$. The assembly work content consists of 7 discrete tasks with times and precedence shown below:

1. Calculate the Takt time ($Takt$) and theoretical minimum number of workstations ($N_{\min}$).
2. Balance the line using the Ranked Positional Weight (RPW) algorithm with cycle time equal to Takt time.
3. Compute the resulting balancing efficiency ($E$) and balance delay ($d$).

Solution:

1. Takt Time and $N_{\min}$:

   • Gross shift duration = $8\text{ hours} \times 60 = 480\text{ minutes} = 28,800\text{ seconds}$.
   • Deduct two 15-minute scheduled line stoppages: $T_a = 480 - 30 = 450\text{ minutes} = 27,000\text{ seconds}$.
   • Takt Time: $Takt = \frac{T_a}{D} = \frac{27,000\text{ seconds}}{300\text{ units}} = 90.0\text{ seconds/unit}$
   • Total work content: $T_{\text{wc}} = \sum_{i=1}^7 t_i = 35 + 50 + 40 + 25 + 45 + 60 + 30 = 285\text{ seconds}$
   • Theoretical minimum workstations: $N_{\min} = \left\lceil \frac{T_{\text{wc}}}{C} \right\rceil = \left\lceil \frac{285}{90} \right\rceil = \lceil 3.167 \rceil = 4\text{ workstations}$

2. RPW Calculation and Ranking:

   • Task 7 followers: none $\implies PW_7 = t_7 = 30\text{ s}$.
   • Task 6 followers: 7 $\implies PW_6 = 60 + 30 = 90\text{ s}$.
   • Task 5 followers: 6, 7 $\implies PW_5 = 45 + 60 + 30 = 135\text{ s}$.
   • Task 4 followers: 6, 7 $\implies PW_4 = 25 + 60 + 30 = 115\text{ s}$.
   • Task 3 followers: 5, 6, 7 $\implies PW_3 = 40 + 45 + 60 + 30 = 175\text{ s}$.
   • Task 2 followers: 4, 6, 7 $\implies PW_2 = 50 + 25 + 60 + 30 = 165\text{ s}$.
   • Task 1 followers: all (2, 3, 4, 5, 6, 7) $\implies PW_1 = 285\text{ s}$.

   Ranking Order: Task 1 (285), Task 3 (175), Task 2 (165), Task 5 (135), Task 4 (115), Task 6 (90), Task 7 (30).

3. Station Assignment (Cycle Time $C = 90\text{ s}$):

   • Station 1 (Capacity = 90 s):
     • Assign Task 1 ($t_1 = 35\text{ s}$, no preds). Remaining time = $90 - 35 = 55\text{ s}$.
     • Evaluate next ranked: Task 3 ($t_3 = 40\text{ s}$, pred 1 done). $40 \le 55$, assign Task 3. Remaining time = $55 - 40 = 15\text{ s}$.
     • Evaluate next: Task 2 ($t_2 = 50 > 15$), Task 5 ($t_5 = 45 > 15$), Task 4 ($t_4 = 25 > 15$). None fit.
     • Station 1 total: Tasks {1, 3}, Station time $S_1 = 35 + 40 = 75\text{ s}$, idle = $15\text{ s}$.
   • Station 2 (Capacity = 90 s):
     • Evaluate highest ranked unassigned: Task 2 ($t_2 = 50\text{ s}$, pred 1 done). Assign Task 2. Remaining = $90 - 50 = 40\text{ s}$.
     • Evaluate next: Task 5 ($t_5 = 45 > 40$). Cannot fit.
     • Evaluate next: Task 4 ($t_4 = 25\text{ s}$, pred 2 done). $25 \le 40$, assign Task 4. Remaining = $40 - 25 = 15\text{ s}$.
     • No other tasks fit in 15 s.
     • Station 2 total: Tasks {2, 4}, Station time $S_2 = 50 + 25 = 75\text{ s}$, idle = $15\text{ s}$.
   • Station 3 (Capacity = 90 s):
     • Evaluate highest ranked unassigned: Task 5 ($t_5 = 45\text{ s}$, pred 3 done). Assign Task 5. Remaining = $90 - 45 = 45\text{ s}$.
     • Evaluate next: Task 6 ($t_6 = 60 > 45$). Cannot fit.
     • Station 3 total: Task {5}, Station time $S_3 = 45\text{ s}$, idle = $45\text{ s}$.
   • Station 4 (Capacity = 90 s):
     • Evaluate next: Task 6 ($t_6 = 60\text{ s}$, preds 4 and 5 done). Assign Task 6. Remaining = $90 - 60 = 30\text{ s}$.
     • Evaluate next: Task 7 ($t_7 = 30\text{ s}$, pred 6 done). Assign Task 7. Remaining = $30 - 30 = 0\text{ s}$.
     • Station 4 total: Tasks {6, 7}, Station time $S_4 = 60 + 30 = 90\text{ s}$, idle = $0\text{ s}$.

   Actual Stations: $N_{\text{actual}} = 4$ workstations. (Matches $N_{\min}$!)

4. Performance Evaluation:

   • Total available capacity: $N_{\text{actual}} \times C = 4 \times 90 = 360\text{ seconds}$.
   • Balancing Efficiency: $E = \frac{\sum t_i}{N_{\text{actual}} \times C} = \frac{285}{360} = 0.7917 = 79.17\%$
   • Balance Delay: $d = 1 - E = 1 - 0.7917 = 0.2083 = 20.83\%$

5. Engineering Conclusion: The line achieves a 79.2% balancing efficiency with 4 workstations, satisfying the 90-second Takt time constraint.

Worked Example 11.2.2: Bottleneck Identification & Little's Law WIP Sizing

Problem: A circuit board assembly line features five sequential automated machines with observed station times: $S_1 = 45\text{ s}$, $S_2 = 72\text{ s}$, $S_3 = 60\text{ s}$, $S_4 = 50\text{ s}$, and $S_5 = 40\text{ s}$. In-line buffer racks between stations hold parts when transfers are delayed. Quality audits indicate an average Work-in-Process inventory of $WIP = 120\text{ circuit boards}$ distributed across the line.

1. Identify the bottleneck station and determine the maximum sustainable line throughput in boards per hour.
2. Using Little's Law, compute the average manufacturing lead time (flow time) for a board traversing the entire line.

Solution:

1. Bottleneck Station & Throughput:
   • The bottleneck is Station 2 because it has the longest processing time: $S_{\max} = S_2 = 72\text{ seconds/board}$.
   • The governing line cycle time is $C = 72\text{ seconds}$.
   • Line Throughput: $TH = \frac{1}{C} = \frac{1\text{ unit}}{72\text{ seconds}} = \frac{3600\text{ seconds/hour}}{72\text{ seconds/unit}} = 50.0\text{ units/hour}$
2. Little's Law Flow Time: $WIP = TH \times CT \quad \implies \quad CT = \frac{WIP}{TH}$ $CT = \frac{120\text{ units}}{50\text{ units/hour}} = 2.40\text{ hours} = 144.0\text{ minutes}$
3. Engineering Conclusion: While the sum of station processing times is only $267\text{ seconds}$ ($4.45\text{ minutes}$), excessive buffer inventory ($WIP = 120$) expands average manufacturing lead time to 2.4 hours ($144\text{ minutes}$).

---

7. NCEES Reference Handbook Tips & Realistic Exam Traps

• Ceiling Function Rounding: Never use standard arithmetic rounding for $N_{\min}$. If $T_{\text{wc}} / C = 4.05$, the theoretical minimum is 5 stations, not 4. A 4-station configuration would be physically incapable of meeting the cycle time without exceeding 100% station utilization.
• Net vs. Gross Operating Time: Takt time problems frequently state an 8-hour shift with scheduled lunch and break times. Always subtract scheduled non-working periods from the numerator. Unpaid lunch is never part of $T_a$, and scheduled paid pauses must also be subtracted if production stops.
• Efficiency Base Trap: When calculating balancing efficiency $E = \sum t_i / (N C)$, always use the actual number of assigned workstations ($N_{\text{actual}}$) and the actual governing cycle time ($C = \max S_j$), NOT the theoretical minimum $N_{\min}$ or Takt time, unless explicitly instructed.
• Precedence Violations in Heuristics: When executing RPW or Largest Candidate heuristics on an exam, never assign a task simply because it has a high weight or large duration. Verify that every immediate predecessor has already been allocated.