Skip to content

waremax sweep

Run a parameter sweep across multiple configurations.

Synopsis

waremax sweep --base <PATH> --sweep <SPEC> --output-dir <PATH> [OPTIONS]

Description

The sweep command systematically varies a parameter across specified values, running multiple simulations to understand how the parameter affects performance. Results are aggregated and ranked.

Options

Required

Option Description
--base Base scenario file
--sweep Sweep specification (see format below)
--output-dir Directory for results

Optional

Option Default Description
--replications 3 Replications per configuration

Sweep Specification

Format: parameter:value1,value2,value3,...

Supported Parameters

Parameter Description
robots Number of robots
order_rate Order arrival rate (orders/hour)
stations Number of stations

Examples

# Sweep robot count
--sweep "robots:5,10,15,20"

# Sweep order rate
--sweep "order_rate:30,60,90,120"

# Sweep station count
--sweep "stations:2,4,6,8"

Examples

Robot count sweep

waremax sweep \
  --base baseline.yaml \
  --sweep "robots:5,10,15,20,25" \
  --replications 5 \
  --output-dir ./robot_sweep

Order rate sweep

waremax sweep \
  --base baseline.yaml \
  --sweep "order_rate:60,90,120,150" \
  --replications 3 \
  --output-dir ./rate_sweep

Station count sweep

waremax sweep \
  --base baseline.yaml \
  --sweep "stations:2,4,6" \
  --replications 3 \
  --output-dir ./station_sweep

Output

Console Output

Running parameter sweep...
  Base scenario: baseline.yaml
  Sweep: robots:5,10,15,20
  Replications: 3
Generated 12 scenarios

Results by Throughput:
------------------------------------------------------------
 1. robots=20                      285.4 ± 12.3 orders/hr
 2. robots=15                      267.8 ± 8.7 orders/hr
 3. robots=10                      198.5 ± 15.2 orders/hr
 4. robots=5                       112.3 ± 6.1 orders/hr

Results saved to: ./robot_sweep/sweep_results.json

Results File

sweep_results.json:

[
  {
    "label": "robots=5_seed=1000",
    "seed": 1000,
    "throughput": 108.5,
    "p95_cycle_time": 72.3,
    "robot_utilization": 0.85,
    "station_utilization": 0.45,
    "duration_ms": 1234
  },
  {
    "label": "robots=5_seed=1001",
    "seed": 1001,
    "throughput": 115.2,
    "p95_cycle_time": 68.9,
    "robot_utilization": 0.82,
    "station_utilization": 0.48,
    "duration_ms": 1198
  }
]

Understanding Results

Ranking

Results are ranked by throughput, showing:

  • Mean throughput across replications
  • Standard deviation

Statistical Significance

  • Higher replications provide more reliable means
  • Large standard deviation indicates high variability
  • Consider using ab-test for statistical comparison

Use Cases

Capacity Planning

# Find optimal robot count for target throughput
waremax sweep \
  --base baseline.yaml \
  --sweep "robots:5,10,15,20,25,30" \
  --replications 5 \
  --output-dir ./capacity_study

Sensitivity Analysis

# Understand impact of order rate changes
waremax sweep \
  --base baseline.yaml \
  --sweep "order_rate:50,75,100,125,150,175,200" \
  --replications 3 \
  --output-dir ./sensitivity

Scaling Study

# Test different warehouse sizes
for size in 5 10 15 20; do
  waremax generate --preset standard --output "size_${size}.yaml" --grid "${size}x${size}"
done

# Sweep robots for each size
for size in 5 10 15 20; do
  waremax sweep \
    --base "size_${size}.yaml" \
    --sweep "robots:5,10,15,20" \
    --output-dir "./scaling_${size}"
done

Best Practices

Replications

  • Use at least 3 replications for general testing
  • Use 5+ replications for important decisions
  • Use 10+ replications for publication-quality results

Parameter Ranges

  • Start with wide ranges
  • Narrow down to interesting regions
  • Include current/baseline values

Analysis

  • Look for diminishing returns (e.g., more robots not helping)
  • Identify knee points where behavior changes
  • Consider interactions between parameters

See Also