Agent Based Simulation Modelling
Agent Based Modelling
The next iteration from Discrete Event Simulation, the Agent Based Modelling approach allows for quick development of complex systems.
In the Agent Based Modelling process you develop the rules and parameters for each individual agent and then the rules on how those agents interact rather than having to define the full system each time you want to add or remove detail to your model.
See your problems before they occur
gemini allows for very complex systems to be defined in a much simpler way, saving time and making the accessibility of high end modelling much simpler to use.
You don’t need to be a computer scientist to be able the build and configure gemini models.
We have a genuine focus around making the software user friendly and with an eye to the kind of software you would be using in your normal day to day activities.
If you have used any kind of CAD style software or mapping tool then you will very easily be able to use gemini.
If you know how to use Google Maps you know how to get started with gemini.
True 3D Environment Simulation
Model your operation in 3D, using your haulage network and infrastructure configuration
With gemini you have the ability to build your model in true space. The data used within the simulation comes directly from your mine planning software with very simple file formats used to allow ease of data transfer.
As well as simple file formats, the data is kept in 3D meaning road gradients are kept, placement of agents is in real space and you have the ability to view your data in several styles – maps like you regularly see in any of the online mapping services such as Google maps or aeriel photography which is also readily available, showing your model on screen in it’s true scale and layout.
When you animate your simulation you will see your trucks changing speed within their environment, slowing down to go up hills, accelerating differently when loaded or unloaded.
A full range of animation controls
Play, Pause, Stop, animation speed
Zoom in and out to animate a specific part of your model
Simple views of your fleet to know what state an agent is currently operating in
Set the style of map you want to use or easily change from one style to another.
Real fleet / equipment Interaction
Each agent in the simulation recognises where it is and where all the other agents are and interacts with them in real space
Detailed Fleet Definitions
Being able to define each piece of equipment individually and then place your equipment in real space you can then see how your fleet interacts with each other.
A truck fleet can be defined with whole of fleet parameters, or each truck can be given it’s own set of parameters, or anything in between. It is a simple exercise to build performance details into your agents including the amount of equipment you have.
It isn’t just trucks and diggers. You can define any type of equipment to work on your haulage network.
As long as there is a path to follow you can define the impacts of other equipment operating on your haulage fleet.
Whether it is a water cart, grader, service vehicles or LV’s traveling on your haulage network you can define them and see what that means to your operation.
Model Fleet Variability
Model your operation in 3D, using your haulage network and infrastructure configuration
All aspects of your simulation data is an agent. Each road, truck, digger, stockpile, loader, grader, water cart and so on is an agent and can contain a range of parameters.
Your roads contain some of those parameters intrinsically as they are in 3D, so length and gradient of each segment is already carried through from the based data. In addition roads can carry speed limits, as well as areas of your network being assigned yield points and stop points as your actual operation does.
Once defined on your road network, each agent travelling on that road will honour its parameters. Set a speed limit and all vehicles travelling on that road will honour it.
Level of Detail
Your mining fleet can be defined with as much detail as you like. You can have it fairly simple where you are only modelling your trucks and digging fleet, or as complicated as you like where you have all other vehicle you might experience on your network defined, such as water carts, graders, light vehicles. See your true network congestion by showing your real day to day activity.
Each piece of equipment, or agent, can carry relevant parameters to allow the simulation to show you as close to real world outcomes as possible. Your agents are assigned variabilities around assumed or actual performance.
Imagine you have truck with a 150t average payload capacity. In the real world you do not get 150t every time the truck is loaded. What you do get is a variation around the mean of that payload. You can define what that variation is so that within the simulation model each time a truck is loaded a payload within your defined distribution is used.
From truck to truck, your payload will vary based on that distribution, but over time your truck payload will be at your mean payload. This results in a much more real reflection of what you might see in your typical hour, day or year.
Source and Destinations
Define links between your source and destination points through a simple list and watch your fleet work between them.
Source and Destination Types
The source and destination points can also be defined in several ways.
Source points can be a list of blocks from within a pit, or can be from a number of stockpiles.
Destination points can be stockpiles, waste dumps or crushers, each with it’s own tipping rules.
For truck and digger combinations you can define the relationship between a source and destination in 2 ways.
You can fix the source and destination. So each time a truck is loaded it will travel to its dump location and then return to the same loading unit each time.
Alternatively, you can have your loading fleet compete for trucks. Once a loading unit has finished loading a truck it will then send out a call asking for an available truck. If you have several loading units working, then they will essentially be competing with each other for the available trucks. This allows you to ensure the usage of your trucking fleet is maximised.
Control your stockpile build and reclaim in great detail.
Identify truck queue easily.
Just like each of your fleet agents your source and destination points are agents and each can carry their own parameters.
Stockpiles can carry a capacity and can contain parameters around how a stockpile is built and then reclaimed. By default, a stockpile follows a build and destroy process, meaning it must be built to capacity before it can be reclaimed. Stockpiles can also be allocated set material to allow blend ratios to be honoured when they are built and reclaimed.
Each individual stockpile can be defined, or combinations of stockpiles combined to help outline practical build and reclaim rules for your simulations.
Practical reclaiming rules can be created around physical space for your stockpiles. If you have 2 full stockpiles next to each other that are both available for reclaim, having a load unit reclaiming from one of those stockpiles in the middle will probably mean you do not have space to add another load unit. Rule can easily be defined to manage this, meaning that the simulations are practical in that the physical distance between fleet is constantly accounted for.
Stockpile Material Definitions
Stockpile can have material definitions created. This forces only the right material to be sent to a stockpile.
The practical nature of this allows you to know if you have enough space for a material, or if sending too much material of a particular type is congesting your haulage system.
Stockpile Build Rules
Do you build your stockpiles in a particular way?
You can create build definitions to show any congestions issues that can arise.
If you must paddock tip the bottom of your stockpile before completing the remainder of the stockpile capacity from a skyway makes a difference to where your trucks travel and can highlight significant congestions issues, especially on an operating ROM.
Define your crusher parameters and see your truck fleet tip, queue as set by your own.
Crusher destination points
The crusher agent can carry a number of parameters such as location, bin capacity, draw down rate, load sensitivity and stockpile combinations.
The combination of the parameters will help determine how often a truck can tip into the crusher. If the crusher bin capacity is small, and the bin draw down rate low, then the combination will lead to truck queue as the truck can only tip once there is capacity within the crusher bin.
This is a very powerful feature within gemini and can provide real practical feedback around your crusher feed rates and infrastructure set points – before you build and operate anything.
Outputs from the simulation allow you to report on your bin levels and draw down rates to easily graph the feed rates and more importantly where your are seeing breaks in feed.
This is an extremely powerful tool in identifying the cause of any feed breaks, either the truck fleet or infrastructure capacity.
As the simulation runs in second, you get the output down to that time breakdown as well.
Crusher destination points can also be assigned an amount of direct tipping that you wish to target.
This works in combination with your truck agents and stockpile agents. As a truck is travelling to your ROM, at a nominated point, the truck analyses the amount of direct tip being targeted and will move to the required destination, either stockpile or crusher to meet the direct tip target amount. In the case of a complicated blending scenario, gemini will manage and target both the direct tip target amount and any stockpile blending ratios that have been defined.
This again means your results give you a very practical and realistic result meaning you can configure your required fleet with confidence.
Test your plans before you execute them – know where your issues are before they are happening. Execute with Confidence
What if Analysis
As it is very quick and easy to both set up a simulation or make changes, testing a plan before you start executing it is very simple, and very valuable.
Being able to identify any opportunity that either helps you reduce the amount of fleet you need to reach your target tonnes, or maximise the tonnes you can mine is an extremely powerful position to be in and is what gemini is able to do – easily and quickly.
Best Possible Shift, Day, Month…..
Moving your loading fleet and trucking fleet into their shift starting locations and then running the simulation against your digging targets, source and destination combinations and required crusher feed can all be run in minutes.
This can then allow you to very quickly adjust your fleet needs accordingly and make the very most of your equipment every shift.
Most importantly, you can identify your potential problems ahead of time and resolve them before they occur.
Use heatmaps to show when and where your fleet interacts with each other.
Due to the very visual nature of gemini, seeing your results and identifying areas for improvement can be very easy.
Displaying a heat map based on truck delays can very quickly help to identify areas on your haulage network that can be improved.
Reporting on truck queue, both at source and destination can show you that you might be over trucked in a particular area.
Watching the animation of your simulation will allow you to easily see if you are getting un-expected truck queue, but then also having the ability to report against the second by second simulation data in tables, graphs or on a map allow for you to see the results the way you want.
Modern, easy to understand user interface with a focus on the user first.
If you are reading this, then you have finished setting up the software !!!!
There is no installation, no downloading anything, no unzipping files.
If you have a modern web browser, say Chrome, you have finished the software installation.
The system is completely web based, you simply log in on your web browser and you have full access to gemini.
Set up your project and then run the simulation. Your local data, such as your haulage network, is then encrypted and sent into the cloud to be processed, and then once complete, returned to you.
This is a huge feature of gemini.
Why do it this way ?
When you process your simulation, you are using high end server processing each and every time. There is no need to have a dedicated high end desktop to run simulations, you simply access this kind of computing power via Google Cloud, Amazon Web Services, etc.
This means you always have access to the most up to date computing power that you don't have to manage or support.
gemini takes care of that for you, and then lets you know once the simulation is finished.
The key component in all this, and one of the best features of this methodology is that you just need a web browser and internet access.
So in case it hasn’t clicked yet – I just need a web browser, and the processing is not done on my computer, but in the cloud – so I can use any device that has a web browser ??
Windows, Apple, Linux – laptop, desktop, iPad, tablet…….any of them
If you have access to a web browser and an internet connection you can run gemini, run your simulations using high powered processing, and all from the device of your choice.