SupplyNetPy

SupplyNetPy is a Python library designed for modeling, simulation, design exploration, and optimization of supply chains and inventory systems. It allows users to create and simulate supply chain networks with various inventory replenishment policies.

Installation

You can install SupplyNetPy using pip.

pip install supplynetpy

Dependencies

• SimPy runs the simulation.
• NetworkX and Matplotlib draw the network.

pip installs all three for you.

Authors

• Tushar Lone GitHub
• Lekshmi P GitHub
• Neha Karanjkar GitHub

Quick Start

Creating supply chain networks

A small network with an infinite supplier, one distributor that reorders with an (s, S) policy, and a customer demand of 10 units every time unit.

import SupplyNetPy.Components as scm

supplier1 = {'ID': 'S1', 'name': 'Supplier1', 'node_type': 'infinite_supplier'}

distributor1 = {'ID': 'D1', 'name': 'Distributor1', 'node_type': 'distributor', 
                'capacity': 150, 'initial_level': 50, 'inventory_holding_cost': 0.2, 
                'replenishment_policy': scm.SSReplenishment, 'policy_param': {'s':100,'S':150},
                'product_buy_price': 100,'product_sell_price': 105}

link1 = {'ID': 'L1', 'source': 'S1', 'sink': 'D1', 'cost': 5, 'lead_time': lambda: 2}

demand1 = {'ID': 'd1', 'name': 'Demand1', 'order_arrival_model': lambda: 1,
            'order_quantity_model': lambda: 10, 'demand_node': 'D1'}

# create a supply chain network
supplychainnet = scm.create_sc_net(nodes=[supplier1, distributor1], links=[link1], demands=[demand1])

# simulate for 20 time units (read here as days)
supplychainnet = scm.simulate_sc_net(supplychainnet, sim_time=20, logging=True)

Modeling what a disruption does to stored inventory

In real life, a disruption is rarely just a pause. A flood can ruin every box in a warehouse, a power cut can spoil refrigerated goods, contamination can force a partial recall, and theft can take part of the inventory. SupplyNetPy lets you describe these physical effects on inventory, in addition to simply marking a node as disrupted.

The basic disruption settings (failure_p, node_disrupt_time) only mark the node as disrupted, so it receives no new orders. They leave its units untouched. To say what actually happens to the units at the node when a disruption starts, set the disruption_impact option as shown below.

# Scenario 1, total loss (for example a warehouse fire or a flood). Destroys all units at the node.
scm.InventoryNode(..., disruption_impact="destroy_all")

# Scenario 2, partial loss (for example a power cut that spoils some refrigerated goods).
# Here, 30% of the inventory level is destroyed each time a disruption starts.
# You can also pass a function instead of 0.3 to randomize the loss each time.
scm.InventoryNode(..., disruption_impact="destroy_fraction",
                  disruption_loss_fraction=0.3)

# Scenario 3, anything more specific (for example contamination of half the units).
# Write a small Python function that describes what to do, and pass it in.
def contaminate(node):
    node.inventory.destroy(amount=node.inventory.level * 0.5,
                           reason="contamination")
scm.InventoryNode(..., disruption_impact=contaminate)

The loss is applied once, at the moment the disruption starts (not repeatedly while it lasts). The units lost and their value are saved on the node as node.stats.destroyed_qty and node.stats.destroyed_value, and the value is automatically subtracted when the simulation calculates the node's profit, so you don't need to do that bookkeeping yourself.

The words used here (inventory level, disruption, customer demand, and so on) are defined in the Glossary.

Documentation

For detailed documentation and advanced usage, please refer to the official documentation.