Case studies

Explore (a sub-section of) my client work. Many more studies are under NDA but this is a good representation of the breadth of my work.

PORT LIBRARY

2023

Context

Full control and ownership of the models were crucial for client success.
Long-term planning of ports requires simulation of traffic at vastly different scales for varying predicted demands.

So far, the client "bought" simulation models but felt the pain of not owning the models and lacking internal capabilities to adjust and improve.

Hence, they decided to develop a library for port model development, so their team can use that going forward.

Approach

The library is very user-friendly but can be extended to any degree.
I developed a comprehensive AnyLogic library tuned to building port models fast, elegantly and reliably.

Within an hour
, you can develop a model for vessel , truck, train, conveyor and pipeline transport for import, export and trans-shipments.

The library has advanced capabilities to detect unsuitable setups, track KPIs and apply different layouts.

Client impact

8 highly detailed port models were developed from the library.
The client built models of all their ports (8), from small simple ports to huge top-tier  global ports (50+ berths).

Models include detailed flows of traffic, predicted weather impacts and toggle varying future layouts.

The client applies these for their strategic investment planning. Importantly, they now fully "own" and know their models. Due to the flexibility of AnyLogic, they can extend the library at any time.

CAR FACTORY LOGISTICS

2022

Context

The plant is 3D-rendered but fully data-driven and packed with visual performance feedback
A large electric-car manufacturer opened a new plant and ramps up production very fast.

Traffic within the plant may become a bottleneck due to space constraints, route setup and "soft factors".

Management wants to "stay ahead" of problems by simulating performance upfront.

Approach

Tons of automated KPIs can be visualized to dig into root-cause analysis
A fully data-driven AnyLogic model re-creates the plant, including assembly lines, tugger routes and part consumption.

I developed a unique spatial setup that allows tugger traffic to be modelled ~100 times faster than traditional traffic models, without loosing critical fidelity.

The model is so generic that it could be be applied to different plants without extra effort.

Client impact

3D heatmaps are applied to visualize various characteristics such as traffic jams, incidents, etc.
The model is used regularly to this day. Both ahead of major design decisions and for small "what would happen if..." queries.

It identifies unexpected bottleneck areas and quantifies the impact of route adjustments, more equipment, different cycle times...

Not least, the model continues to be improved by the internal simulation team, after training and upskilling by me.

VACCINE LOGISTICS SUPPORT

2022

Context

The 3D model is fully interactive and visualizes operations and performance.
A vaccine manufacturer experienced explosive growth and struggled to keep up with production. The main bottleneck was the physical limitations of the manufacturing plant itself.

They asked IPOL GmbH to help them improve processes and maximize throughput within the physical limitations.

Approach

A 100% data-driven setup allows rapid model setup. You can build the most complex sites with intuitive inputs.
IPOL GmbH envisioned a generic simulation tool that allows automated creation of any client physical layout: sites, buildings, elevators, etc. These would be filled with pedestrians navigating with custom route finding.

We co-developed a powerful AnyLogic model that creates such models from data.

Client impact

Even the dashboards are data-driven: No matter your data complexity, the KPIs will cater for it.
IPOL experts learned advanced AnyLogic modelling techniques while serving their client at the same time. 
IPOL plans to re-use the setup for future clients as it is generic enough to be applied for different industries. eduling.

PROJECT PORTFOLIO DIGITAL TWIN

2021 - 2022

Context

Single runs display detailed performance KPIs
Goldratt Research Labs is a thought leader on “Theory of Constraints”.

They decided to apply their expertise not only in manufacturing and supply chains but bring it to project management, specifically critical-chain project management.

Approach

Several scenarios can be compared directly.
We designed a highly functional, yet intuitive tool to load any project portfolio.
Users can analyze their current situation and simulate the future, testing different policies with real-world uncertainties.

Client impact

Best-in-class custom charts were developed for this tool.
This tool is the most comprehensive way to plan and test your company’s projects. It has features not seen in comparable tools, specifically its ability to actually simulate future uncertainty.

SEMI-CONDUCTOR SCHEDULING OPTIMIZATION

Global semi-conductor manufacturer
Winter 2020-21

Context

Semi-conductor manufacturing is extremely challenging.
Manufacturing semi-conductors is incredibly complex, requiring dozens of steps on just as many machines.

The client relied on traditional Excel-based scheduling driven by human experience alone.

Approach

This is only indicative, real model worked without animations.
I developed a highly detailed digital twin of client operations while retaining high-performance to enable optimization of schedules.
This required very advanced modelling techniques with low-level coding and assemblies.

Client impact

Individual wafers got their custom schedule derived from an optimization (image only indicative)
The initial prototype convinced the client to invest into a very large study to turn this into the new scheduling tool.

I pulled in additional resources to kickstart full development. The product is being used every day since it has been developed.

COVID-19 FACILITY SIMULATION TOOL

Summer 2020

Context

Intuitive GUI tells a story
During the Covid-19 pandemic of 2020, many offices and factories were forced to re-structure their processes around social-distancing.
The Silicon-Valley startup Yogeo asked me to create a generic tool to help solve those issues.

Approach

Dynamic 3D animations show problem areas.
I created a “quasi” library of typical blocks like spaces, desks, toilets or cafeterias.
Users can drag these in themselves and link them with spatial elements to re-create any real setup within minutes.
This is a unique approach combining logic and animation in AnyLogic.

Client impact

Check this flyer with more info
Yogeo applied the tool to discuss implementations with clients.
They also used the library elements to adjust my demo layout for their own client needs, proving the simplicity it provides.
To this day, the model is one of the most successful models on the AnyLogic Cloud.
“Business leaders and creators in the real world will find Benjamin’s modelling skills to be invaluable in shaping their future.”
Sri Jagannathan, CEO Yogeo

PRODUCT PORTFOLIO DIGITAL TWIN

Tier-1 Aerospace Supplier, France
Spring 2020

Context

  • Aerospace super-alloy manufacturer with complex product portfolio
  • Client wanted to evaluate impact of future (simplified) product portfolio on cost, revenue and machine utilization
  • 10000+ different SKUs with very complex flows across several plants

Approach

  • Data-driven, self-configuring model of all plants and machines
  • SKUs follow required production flows and accrue costs and revenue
  • Easy setup of new product portfolio scenarios
  • Innovative data visualization

Client impact

  • First time client was able to see actual product flows for any group across any machine
  • Recommended several future portfolios with respective impact on cost and revenues

MATERIAL DESIGN: UI LIBRARY FOR ANYLOGIC

2019-2022

Context

Smooth animations, like an App
The AnyLogic simulation tool offers great simulation capabilties. However, the default model user interface (UI) is often old-fashioned and not intuitive.

Why not move to UI that we are all familiar with, i.e. modern app design?

Approach

Applying modern Material-Design principles
Together with Goldratt Research Labs, we decided that great models benefit from good UI.

We developed the first commercial AnyLogic library where users can create amazing UI within minutes that look and feel like modern apps.

Client impact

Enabling beautiful dashboard
Goldratt Research Labs now develops all their simulation tools with our Material Design library.

Other customers around the world use it every day and shorten their model design time while improving visuals massively.
“The library is very intuitive, easy to learn and use, and very well documented. The accompanied course on Udemy was awesome. You can get up to speed to build models with super attractive user interfaces. Thanks for producing valuable contents and sharing them with community. It has made a difference in our work.”
Davood Astaraky, Ottawa Hospital Research Institute

INVENTORY MANAGEMENT DIGITAL TWIN

Since 2019

Context

Write your awesome label here.
Intro video
Goldratt Research Labs is a leading provider in helping clients to make “better faster decisions when it really matters”.

For their clients, they use simulation models extensively to train insights into inventory management policies. The goal is always to reduce shortages and surpluses in retail.

Approach

Easy policy comparison and powerful KPIs
Based on Eli Goldratt’s famous book “Isn’t it obvious”, we built a digital twin of a retail store and a distribution center.

Uniquely, the simulation “tells a story” and takes the user through a learning journey, also applying intuitive user interfaces and charts.

Client impact

Detailed insights into inventory performance
The model is used extensively in client training to educate about the advantages of “theory-of-constraints” inventory management over traditional approaches.

Recently, the model was extended to include DDMPR policies as well. Moreover, it is now a full digital-twin that clients can apply to real data and situations.

ALUMINIUM PLANT DIGITAL TWINS

Global aluminum producer (USA & Europe)
2018-2019

Context

The entire plants were modelled
Two client flat-rolled aluminium plants faced dynamic bottlenecks in North America and Europe. This was caused by several issues that could often not be identified.

Several improvement ideas existed but it was impossible to analyse which one would be the best.

Approach

Detailed dashboards allow immediate insights
I designed a full “digital twin” simulation model of each plant, modelling individual coil flow through all machines.

After calibrating for 95% accuracy on historic data, we tested several scenarios to improve throughput and OTIF.

Client impact

Even coil temperatures are modelled and animated
The simulation model was the key tool to decide between scenarios.

The chosen scenarios lead to an increase of throughput by 10%, a WIP reduction of over 5% and a lead-time improvement of over 1 day.

STEEL-PIPE SUPPLY CHAIN

European steel-pipe manufacturer
Spring 2018

Context

Summary KPIs
  • Client manufacturer has several sites across Europe, each managing production steps for steel pipes (rolling, heat treatment, threading…)
  • Current shipping is planned ad-hoc, creating large overhead
  • Impossible to optimize due to complexity
  • Simulation could help test new planning rules

Approach

Dashboard comparison
  • Create AnyLogic model re-creating production sites, current shipping strategies and cost structure
  • Train the trainer” approach: taught client resource in simulation modelling so he could help with model development from week 3
  • Initial results created within 2 weeks, refined iteratively

Client impact

GIS-map to visualize actual supply chain
  • Tested 8 policies and ranked by various goals
  • Ranking led to further refinement of initial policy ideas. Those were retested in the model again
  • Left client capable of further model work and developing new models for different problems

GOLD MINE DIGITAL TWIN

Argentina
2017

Context

Fully animated open-pit mine
  • Sub-optimal performance due to shifting bottlenecks
  • Client did not understand why bottlenecks appear. Classic OEE analysis cannot help here
  • Client developed improvement ideas but could not predict impact

Approach

Custom-design dashboards
  • A highly accurate mine model was designed using AnyLogic
  • Including all assets (trucks, drivers, drills, crushers…), failure modes
  • Uniquely, we modelled the ore and leachpad operations as well using the fluid library and chemical reactions

Client impact

Fluid simulation of leach-pad reactions
  • 12 man-week effort led to a highly accurate model (reproduced the past within 5%)
  • Identified up to 1m tons of material loss reduction
  • Predicted unexpected future bottlenecks
  • Daily operational use due to self-explanatory UI design

S & OP PLANNING

Large consultancy
2017

Context

Supply chain visualized on a GIS map
  • S&OP (Sales and operations planning) is often done without a full understanding of the system dynamics
  • Educating S&OP professionals about unexpected problems is very important
  • Serious games” can help by experiencing a virtual S&OP environment

Approach

Individual factories modelled in the same tool
  • We built a “digital twin” of a lemonade business with suppliers, factories and DCs across Europe
  • Uniquely, the model recreates individual factory dynamics down to individual bottles
  • Data-driven setup that users can easily adjust at runtime

Client impact

Summary KPIs to see impact of changes
  • Used regularly for training of clients of a large management consulting firm
  • Extended existing (real) “serious game” of a lemonade factory with a virtual S&OP component
  • Users report on the unexpected and fun aspect of training for S&OP end-to-end

AIRCRAFT-ENGINE-FLEET DIGITAL TWIN

Engine manufacturer, UK
2013-16

Context

 
Our client leases a big chunk of their engines to airlines, instead of selling them. All risk on downtimes and repairs is on the manufacturer.

Therefore, predicting downtimes and managing repair shop availability is crucial to their business model. A strategic and live digital twin of the aircraft fleet can help.

Approach

 
We created one of the most complex AnyLogic models at the time, modelling aircraft engines down to individual screws. The model was fed from live and historic data.
Part failures could cause cascades of errors and predictive maintenance became feasible for the first time

Client impact

 
The tool became an integrated part of the manufacturer’s live operations.

It is still applied every day to predict on short- and long-term fleet performance.
It has helped to shape major investment decisions on overhaul facilities.

REME MAINTENANCE PLANNING

UK Ministry of Defense
2014 - 2015

Context

“REME” stands for the Royal Electrical and Mechanical Engineers
The UK Ministry of Defense (MoD) sends helicopter maintenance personnel to crisis areas along with the pilots. Naturally, this creates a lot of stress, leading to a high rate of people leaving out of frustration.

Given the high cost of training, the MoD wanted to see if better crew handling could improve morale.

Approach

Individual agents arranged in novel circular layout to show KPIs intuitively
We created an agent-based model of maintenance personnel, helicopters, training exercises and deployments.

Personnel agents had an internal “psychology” state model of their mood, influenced by their day-to-day activities.

Client impact

Visual feedback provides direct information on performance
We showed that rate of personnel leaving could be reduced substantially by slicing “nights away from home” into shorter segments.

The client applied the model to investigate real training schedules and upcoming deployments to apply improved scheduling.
Created with