In this case study, Tolgahan Tarkan from Dijitalis presents how he optimized AGV fleet investments in large production system and saved more than 1.5 million USD  in capex using Simio simulation software. The presention was broadcasted at the Simio Sync 2025 simulation modeling summit.

The consultant shows details of how he built the model quickly using auto-create feature of Simio. The simulation model is highly flexible since most of the model is created using data-driven methodology therefore enabling any data in the data tables to be imported from Excel such as production schedule, product routings, bill of materials (BOM) so on.

The simulation model also acts as production schedule verification tool in terms of AGV/AMR timeliness. Dijitalis consultant also shows two heat maps, one for aisle congestion and second for aisle traffic flow rate all using Simio simulation software. The Gannt charts in Simio also provides information about delay causes in the initial system. Therefore it is possible to do root cause analysis with the Gannt charts of Simio software.

The consultant shows details of how he built the model quickly using auto-create feature of Simio. The simulation model is highly flexible since most of the model is created using data-driven methodology therefore enabling any data in the data tables to be imported from Excel such as production schedule, product routings, bill of materials (BOM) so on.

The simulation model also acts as production schedule verification tool in terms of AGV/AMR timeliness. Dijitalis consultant also shows two heat maps, one for aisle congestion and second for aisle traffic flow rate all using Simio simulation software. The Gannt charts in Simio also provides information about delay causes in the initial system. Therefore it is possible to do root cause analysis with the Gannt charts of Simio software.

Using PuLP this Python template solves a capacity planning problem using linear programming. 

Linear optimization in Python for oil pump capacity planning

In the exemplary case study solved by this Python model a new oil field is prepared. For this, oil pumps must be purchased. Different pump types have different production capacities and purchasing prices.

The problem is to minimize purchasing expenses while at least securing required production volumes.

Another constraint is the available surface area of the oil field. Above figure displays an exemplary oil field in Texas.

KPIs traced by the linear capacity planning Python model

The optimization model considers the following KPIs:

• Purchasing expenses
• Production output
• Surface area occupied on the oil field

Who will benefit from this linear programming template?

Linear programming is a fundamental mathematical programming technique, applied not only in capacity planning. For example, logistics networks can be optimized with linear programming as well. It is also used for e.g. marketing campaign planning, pricing, and much more.

Hence, getting familiar with linear programming and owning a Python template for implementing a linear optimization program in Python, can be beneficial to a wide group of users. This includes students, analysts, and managers – in manufacturing, logistics, purchasing, accounting, controlling, and marketing.

More about linear optimization in Python and other programming languages

If you are interested in linear programming, capacity planning, and mathematical optimization, here are some related articles that might help in getting you started:

• Optimization via master production scheduling
• Price and inventory optimization
• Integer programming
• Linear programming in Julia
• Linear program with Google ortools in Python

A case study is included by the downloadable product. The case study demonstrates barge transport simulation for marine shipping on the Mississippi River. The case study comprises a problem description and exemplary model implementation in Python. The model implementation consumes the framework and solves the case study.

In short this downloadable Python project comprises:

• Framework in Python for modelling and simulating barge transports.
• Detailed case study description of barge transport between Cairo and New Orleans.
• Example model consuming the framework and solving the case study.
• Configuration file with relevant parameters related to the case study.
• 1-hour video call for model training and/or adjust with the model developer.

The framework uses SimPy and implements a discrete-event simulation approach. The framework tracks and visualizes relevant KPIs.

Who will benefit from this barge transport SimPy model?

This downloadable simulation product suits supply chain analysts, operations researchers, transport planners, and students that want to:

• Learn about discrete-event simulation in Python (SimPy).
• Learn about barge transport planning and operation.
• Learn how to model barge transport processes in Python (SimPy).
• Improve existing barge transport processes and projects.

This virtual product includes 1 hour of customizable training, in the form of a video call. This training takes place in a way that best suits your needs.

You can learn more about SimPy in e.g. the following articles:

• https://www.supplychaindataanalytics.com/visualizing-simpy-job-shop-simulation-data/
• https://www.supplychaindataanalytics.com/job-shop-simulation-in-simpy/
• https://www.supplychaindataanalytics.com/visualizing-stats-with-salabim-des-python/

Example models, consuming the framework, included in this downloadable virtual product comprise e.g. the following:

• Agent-based SIR model implementation for disease spread
• Agent-based word of mouth model
• Agent-based social segregation model
• Conway’s game of life

The product includes 45 min free consultation that can be used in one or two short video calls. In these video calls any questions that you might have, related to the agent-based framework purchased here, can be discussed and clarified. The time can also be used to get a quick training in how to use the framework.

The product is downloadable 1 time. If you need another download you can contact us via the contact form. The download limit is in place to ensure that the product is only downloaded by one and not several users.

The download link is accessible for 30 days. If you need a download link renewal please contact us via the contact form.

Currently no tax is collected on this items, as per existing regulation and revenue limit of this website. In future, taxes might be collected and in that case also specified in the invoice for tax-deduction purposes, if our services exceed currently regulated revenue limits.

In detail this downloadable SimPy simulation framework comprises:

• SimPy object-oriented modeling library for modeling poultry supply chain
• Example model implementing exemplary poultry supply chain 
• Conceptual model description
• Configuration file
• Integrated statistics generation

To learn more about the various different simulation methods you can e.g. read this simulation method introduction.

Poultry supply chain elements included in the model

The framework comprises the following supply chain entity models: 

• Egg supplier
• Hatchery
• Brooder
• Farmer coops
• Slaughterhouse
• Meat processing

Information flow assumed by the simulation framework

The exemplary poultry supply chain model included by this virtual product implements/assumes a PUSH controlled supply chain. The egg supplier is the source of the material flow, and the meat processor is the sink. Egg shipments from the egg supplier are supplied to the hatchery upon order placement by the hatchery, using a s,S-based inventory control logic. From there on material flow is PUSH based, all the way to the slaugtherhouse’s finished product stock (carcasses). Meat processing, facilitating the sink in this example model, pulls its demand from the slaughterhouses’ inventory. For more details read the conceptual supply chain description (part of the product).

Relevant poultry  growth assumptions and constraints

This framework focuses on supply chain simulation. It thus makes some simplifying assumptions. For example, chicken growth dynamics are not modelled in detail. Instead  the model assumes process dwell times and chicken start and end weights for relevant processes.

This applies to the following processes:

• Broodery. The framework, via its configuration files, allows for brooding time specification. It is assumed that all hatched undergo the same brooding time. It is furthermore assumes, that all chicken have the same weight when exiting the broodery.
• Farm coops. The framework allows two different population types to be modelled. For example, hens and toms. Coop dwell is the same for all chicken and all populations of the same type. Dwell times in the coop can however be different for the two different population types. This allows to reflect e.g. faster growth of one population type over another. All chicken, regardless of their type, are assumed to have the same weight at the end of their coop dwell time. This is the “slaughter-ready weight”.

KPIs calculated by this framework

This framework calculates and tracks the number of units in a process or in stock (inventory). The same is true for final demand at the end customer (meat processing facility). 

Modelling demand and inventory becomes unit neutral in this way. That allows for straight forward specifcation of mortality rates (referred to as “scrap” as per production industry terminology) and unsuccessful hatching ratios (failure ratios). Morever, chicken weights at the various stages of the supply chain can thereby easily be assumed and multiplied with the quanitites. The same is true for costs, e.g. purchasing costs, feeding costs (per period and chicken unit), holding costs, processing costs, etc. 

It must be clearly understood that a cost model is not part of this simulation model. But since stocks and flows are measured in the same unit (“number of chicken”, “number of eggs”, “number of carcasses”) a cost model can flexibly and easily be applied on top of the simulation results generated by this framework.

Exemplary model output 

Here are some exemplary statistics and statistics visualizations generated by the SimPy framework comprised by this virtual product:

Textbook

Model Capabilities

This model analyzes the pricing strategies and car class upgrade decisions in a complex car rental system to maximize profit. The simulation model embodies a ‘reservation system’ which incorporates the logic of entities requiring a resource in the future time of simulation time.

The model is built with Extendsim 9.3 AT full version (commercial license) and it is compatible with Extendsim 9.3 LT version as well as the model size is less than 75 blocks.

The car rental simulation model built in Extendsim incorporates vast flexibility and can be fitted to a car rental system of any size. The model has no limitation on the number of locations and car classes (categories) in the system. The model has a simple interface where users can copy and paste car rental data from excel in the model. The model can work in 3 different modes:

• Read car rental data from excel and generate demand accordingly
• Create car rental data based on random distributions for
  • Customer arrival time,
  • Rental duration (drop-off time)
  • Difference of time between reservation and rental start, (Customer Type: walk-in or reservation)
  • Percentage of customers requesting each car category
  • Pick-up location preference percentage
  • Drop-off location preference percentage
• Read car rental data inputted from excel and generate random additional car rental data.

The simulation model also makes it is to possible to decide on the number of upgrades to be offered to customers if the car class they asked for is not available at the pick-up time. The number of upgrades to be offered to the customers can be inputted by clicking the upgrades button at the main user interface.

The user can also decide on the precision on the calculated mean values of the turnover. The simulation model runs replications until relative error for the calculated mean of the turnover is less than the value inputted by the user.

Case Description

There are two types of customers who wish to rent a car; walk-in and reservation customers. Walk-in customers arrive to a car rental location where they ask for the availability for car category they wish the rent, for the duration they desire and to be dropped off at their desired destination. Walk-in customers ask for the immediate rental of car they wish to rent for. The reservation customers, however, make a reservation for the car class they wish to rent for the rental starting date in the near future and duration they desire, and they pay or secure the amount to the car rental company at the time of reservation.

All demand (walk-in and reservation customers) arriving to the car rental office or reservation system, has the following information:

• The class of car the customer is willing to rent
• The duration of car rental (in days)
• The maximum amount a customer willing to pay per day for renting their desired car class
• The location for the start of the rental
• The location for the end of the rental

The walk-in customers are not aware of the availability of the cars and its price before arriving to the car rental office. They make decisions based on the response they get regarding the availability and the price of the cars. Reservation customers entering the reservation system act in a similar fashion regarding the price and availability of the cars. If they find a car available within their budget and for time and duration they require, they fill in the online form to complete the reservation. In this case study the budget (maximum amount a customer is willing to pay) for a specific car class is determined by assigning exponential distribution for this value. This is because, exponential distribution is the best fitting distribution to represent the maximum amount a customer can pay for any product or service. This feature of the simulation model adds real-life complexity to the system where the demand for any car class is dependent upon the price of that car class.

The customers arriving to car rental system already know the car class they want to rent for which can be determined either in the inputted demand data or by percentage values for each car class in the random demand generation mode. In this case study simulation runs with the inputted demand data where all car classes have uniform demand. The initial number of cars available at each location for each car class is set to 5. However, by changing the price for any of the car classes the demand for that car class will also change which will affect the availability of the car class. Further, by enabling upgrades to customers the demand for next (higher) car class will also be affected. The rental fee (turnover) is calculated by the following formula: the number of rental days * price for the asked car category. In this case study, the effect of different pricing strategies will be investigated. Further, the effect of offering upgrades on the turnover will also be analyzed and commented upon.