[Challenge #02 – Analysis] How Many AGVs Are Enough?

🧪 [Challenge #02 – Analysis] How Many AGVs Are Enough?

A simulation-driven search for the balance between productivity and efficiency

---

⚙️ Modeling Process Overview

1️⃣ Layout Setup & Base Configuration

• Flow: Start → B zone (Station B) → A zone (Station A) → End
• AGV Routing:
  • Each station is paired with a Marker for pickup/drop-off.
  • AGVs follow a closed-loop route via Track/Line objects.
  • AGVs return to the AGVPool after completing each job.
• AMR Configuration:
  • Turning radius: 0m (pivot rotation)
  • Capacity: 1
  • Receives only one command per trip
  • Always returns to AGVPool after the job

---

2️⃣ Logic Development

• Job generation:
  • Source generates jobs when downstream stations in B are available.
  • Jobs are pushed to a sequence queue based on availability (B_1, B_2, B_3).

--Source exit ctrl

waituntil B_1.set = false or  B_2.set = false or  B_3.set = false -- Waituntil B1 Station Empty

if B_1.set = false
	B_1.set := true
elseif B_2.set = false
	B_2.set := true
else
	B_3.set := true
end

seq.push(?.name) -- Set Seq queue
seq.push("")

 

waituntil A_1.set = false or  A_2.set = false  or  A_3.set = false -- B only

seq.push(?.name)
seq.push("")

• Execution:
  • AGV waits in AGVPool until a job is ready.
  • Based on the command string, AGV picks up the MU and moves it to the corresponding station.
  • When the station in B completes processing, the MU is moved to the next available A zone station.
  • After unloading, the AGV always returns to the AGVPool.

-- param attribute: string, oldValue, newValue: string
-- param attribute: string, oldValue: string
param newValue: string

var A :object

waituntil AGVPool.NumIdleAGVs > 0
var agv := agvPool.getIdleAGV
seq.pop
switch newValue
case ""
case "Source"
	agv.destination := O_M
	waituntil agv.destinationwasreached
	wait loadunloadTIme
	source.mu.move(agv)
	if B_1.empty = true
		agv.destination := B_1_M
		waituntil agv.destinationwasreached
		wait loadunloadTIme
		agv.mu.move(B_1)
	elseif B_2.empty = true
		agv.destination := B_2_M
		waituntil agv.destinationwasreached
		wait loadunloadTIme
		agv.mu.move(B_2)
	else
		agv.destination := B_3_M
		waituntil agv.destinationwasreached
		wait loadunloadTIme
		agv.mu.move(B_3)
	end
	agv.destination := AGVPool
	waituntil agv.destinationwasreached
case "A_1"
	agv.destination := A_1_M
	waituntil agv.destinationwasreached
	wait loadunloadTIme
	A_1.mu.move(agv)
	agv.destination := Marker
	waituntil agv.destinationwasreached
	wait loadunloadTIme
	agv.mu.move(Conveyor)
	agv.destination := AGVPool
	waituntil agv.destinationwasreached
	A_1.set := false
case "A_2"
	agv.destination := A_2_M
	waituntil agv.destinationwasreached
	wait loadunloadTIme
	A_2.mu.move(agv)
	agv.destination := Marker
	waituntil agv.destinationwasreached
	wait loadunloadTIme
	agv.mu.move(Conveyor)
	agv.destination := AGVPool
	waituntil agv.destinationwasreached
	A_2.set := false
case "A_3"
	agv.destination := A_3_M
	waituntil agv.destinationwasreached
	wait loadunloadTIme
	A_3.mu.move(agv)
	agv.destination := Marker
	waituntil agv.destinationwasreached
	wait loadunloadTIme
	agv.mu.move(Conveyor)
	agv.destination := AGVPool
	waituntil agv.destinationwasreached
	A_3.set := false
case "B_1"
	if A_1.set = false
		A_1.set := True
		A := A_1
	elseif A_2.set = false
		A_2.set := True
		A := A_2
	else
		A_3.set := True
		A := A_3
	end

	agv.destination := B_1_M
	waituntil agv.destinationwasreached
	wait loadunloadTIme
	B_1.mu.move(agv)
	if A = A_1
		agv.destination := A_1_M
		waituntil agv.destinationwasreached
		wait loadunloadTIme
		agv.mu.move(A_1)
	elseif A = A_2
		A_2.set := True
		agv.destination := A_2_M
		waituntil agv.destinationwasreached
		wait loadunloadTIme
		agv.mu.move(A_2)
	else
		A_3.set := True
		agv.destination := A_3_M
		waituntil agv.destinationwasreached
		wait loadunloadTIme
		agv.mu.move(A_3)
	end
	B_1.set := false
	agv.destination := AGVPool
	waituntil agv.destinationwasreached
case "B_2"
	if A_1.set = false
		A_1.set := True
		A := A_1
	elseif A_2.set = false
		A_2.set := True
		A := A_2
	else
		A_3.set := True
		A := A_3
	end

	agv.destination := B_2_M
	waituntil agv.destinationwasreached
	wait loadunloadTIme
	B_2.mu.move(agv)
	if A = A_1
		A_1.set := True
		agv.destination := A_1_M
		waituntil agv.destinationwasreached
		wait loadunloadTIme
		agv.mu.move(A_1)
	elseif A = A_2
		A_2.set := True
		agv.destination := A_2_M
		waituntil agv.destinationwasreached
		wait loadunloadTIme
		agv.mu.move(A_2)
	else
		A_3.set := True
		agv.destination := A_3_M
		waituntil agv.destinationwasreached
		wait loadunloadTIme
		agv.mu.move(A_3)
	end
	B_2.set := false
	agv.destination := AGVPool
	waituntil agv.destinationwasreached
case "B_3"

	if A_1.set = false
		A_1.set := True
		A := A_1
	elseif A_2.set = false
		A_2.set := True
		A := A_2
	else
		A_3.set := True
		A := A_3
	end

	agv.destination := B_3_M
	waituntil agv.destinationwasreached
	wait loadunloadTIme
	B_3.mu.move(agv)
	if A = A_1
		A_1.set := True
		agv.destination := A_1_M
		waituntil agv.destinationwasreached
		wait loadunloadTIme
		agv.mu.move(A_1)
	elseif A = A_2
		A_2.set := True
		agv.destination := A_2_M
		waituntil agv.destinationwasreached
		wait loadunloadTIme
		agv.mu.move(A_2)
	else
		A_3.set := True
		agv.destination := A_3_M
		waituntil agv.destinationwasreached
		wait loadunloadTIme
		agv.mu.move(A_3)
	end
	B_3.set := false
	agv.destination := AGVPool
	waituntil agv.destinationwasreached
end

agv.isidle := true

• Important Behavior:
  • Job sequencing and state flags (set := true/false) are used to reserve stations.
  • AGV logic ensures MUs flow only when downstream space is available.

---

3️⃣ Experiment Design

• Tool: Experiment Manager
• Independent Variable: Number of AGVs (1 to 5)
• Fixed Conditions:
  • Processing time: 60 sec (both B and A zones)
  • Simulation period: 7 days
  • Source interval: fixed
• Key Metrics:
  • Daily production (Units Per Day - UPD)
  • AGV Idle Time
  • Station Waiting Time
  • AGV Utilization

---

📊 Simulation Summary

AGVsDaily Output (UPD)Comment

1	225	Severe bottleneck
2	433	Improved, but still under target
3	645	✅ Target achieved
4	623	✅ Slight drop, higher idle time
5	728	✅ Max capacity reached, some idle overload

 

Note: Results may vary depending on actual simulation conditions.

---

🔍 Key Insights

• With 1 AGV, a clear bottleneck and long waiting times occur.
• 2 AGVs significantly improve flow, but still fall short of the goal.
• 3 AGVs achieve the daily target of 600 units with balanced performance.
• 4 AGVs lead to a slight increase in idle time, with limited production gain.
• 5 AGVs reach the system’s practical maximum capacity, but show diminishing returns and increased inefficiency.

---

📌 Conclusion

• ✅ Optimal number of AGVs: 3
• This configuration strikes the best balance between investment and performance.
• Deploying more than 3 AGVs results in marginal gains and higher idle time.
• If maximum capacity is the business priority, 5 AGVs may be justified — but with clear trade-offs.