Why do mining companies do data so wrong?
Here’s the main areas that I’m going to cover as I roll through my perspective on why mining companies do data so wrong.
Data collection
Cleansing data
Data validation
Data mapping
Data warehousing
Data processing
Timing of data
Presentation of data
Format data
data visualisation
Supporting performance through data
Knowing that data will never tell the entire story
Focus on the metrics that matter
live and die by your truth set
automate the data and reports
So that just gives a bit of an overview on the key areas that I’m going to hit in this article.
So let’s start with data collection
Mining data collection, to me is one of the foundations and fundamental areas where mining companies get it so wrong.
Just because we have a LOT of data, doesn’t mean we have good data.
It is in the collection of data and the alignment of that data that enables us to do something great.
In my opinion data collection comes down to being able to measure all of the things that really matter.
But how do you know what it is that you are going to need.
Sometimes you don’t know what’s going to matter until you start diving into some of that operational mining data.
But that’s a question and a conversation for another time.
So when you’re collecting data, you need to be looking at how that data is going to influence the decision making process.
How do you enable your mining operators, maintainers, supervisors, superintendents, engineers, and other specialist roles on site to support the operation to improve their daily performance?
Mining Industry Data collection starts at the source
The closer you can get to the source while ensuring that the data is collected at the highest possible rate while also ensuring that it is accurate.
While doing this you also need to make sure that you have a universal timestamp that is well understood.
This sounds super easy but in the world of Mining trucks, Excavators, plants in their 40’s 50’s and beyond that is not always that easy.
But by focusing adequate time and attention on accurate data collection it enables you to build a much bigger and better picture of what’s taking place in your mining operation.
Cleansing of your mining data and why it is so important
Do you have a clear process that is automated to cleanse your operational mining data?
Is that same system designed to support the identification of areas for improvement in the data collection.
Before you go too much further you (or your team) ARE going to be making the wrong decisions based on the data that you have Infront of you. Unless of course you have a great cleansing strategy, a curious and engaged team and a continuous improvement culture.
I cannot stress this highly enough.
If you are not cleansing your data effectively and identifying areas to improve your data collection, you will likely be poorly informed.
It’s super important that you’ve got a clear method to cleansing and improving your data collection process that is automated and enables the operation to move forward.
Mining Data Validation Why It Is So Critical
I learnt this early in my career all the way back when I was in HR and was told in no uncertain terms that the reports that I had in front of me did not reflect reality on the crew.
Fortunately I found this out by going to ask the team where this data was being pulled from. For some reason it just didn’t seem right.
So, I went to ask…
As it turned out I was fortunate that I went and asked!
The reason that the data was wrong is that the process had changed and the data and options had not been mapped.
This meant that the team that was dealing with the systems were just trying to do their best (while getting frustrated).
Data validation is generally not that difficult in most instances it is a case of going, looking, observing and asking questions but it can get a little bit more complex.
Couple of Tips In Mining Data Validation
Go to the source of data collection
A simple stop watch can be super handy
Using a camera that has slow motion play back can be super handy
Drones can also be awesome for a birds eye view
Take multiple samples and different times
Ask questions and listen (You have 2 ears and 1 mouth use them in that ratio)
But what about a complex mineral processing plant?
What if there are multiple processes running at the same time?
You really do have to be persistent at on going back a number of times to see what’s happening out there.
Also don’t try to validate large areas of a process at once.
If you are able to use some mining process subject matter experts otherwise known as operators to map out your value driver tree and able to understand relationships.
Once you have mapped this out then start on what you and the mining subject matter experts believe may be the most critical paths in the Value Driver Tree (VDT).
Once you know the critical paths got and check/validate each of the nodes in that critical path starting as deep in the process as possible.
Complex processes may not be easy but when they are broken down they can be made a little more simple.
This may not always be the case but you have to find methods to validate the data at each of the data sources to ensure that it’s painting the right picture.
Not doing this effectively may also lead to you and your team making the wrong decisions and heading down the wrong path for an extended period of time.
What about data mapping?
So often, we don’t understand the connection between our data sources and just what does and doesn’t impact on other streams flowing down the value driver tree.
It’s important that you g through the process of mapping your data!
In my opinion it is super critical to ensuring that as you’re going through the collection, cleansing, validating and processing in the right way.
Enabling you to understand the connection between your data sources, and that you’re able to clearly identify when a particular component in your process isn’t performing effectively.
Using your data mapping to solve and answer questions
Is it a data collection issue?
Is it a sensor issue?
Is it something else altogether?
Is it human impute issues?
If it is human related is there a system to support them that is clear and concise?
Data Warehousing
About 10 years ago, I heard this concept of a data lake and in reality, I really started to wonder what they meant by a lake.
For me, the stories of Loch Ness monsters and other creatures lurking beneath was the first thing that came up.
If you’ve got all of these monsters hidden away in the dark areas of your lake, then do you really have a true picture?
Are you creating ambiguity across your organization because no one truly understands what that looks like.
I think that might be the case for a lot of mining companies out there. And yes I have seen A lot of mining companies in action all across the globe.
What about in the shallows of this lake, it’s muddy and murky with sticks jutting out all over the place no one wants to go in there!
There are also those sections that just look scary that no one knows what is just beneath the surface and all these and other bits and pieces really do impact your data validity in a in a meaningful way.
Operational Mining Data Processing – “I need my data live“
If I had a dollar every time I heard this in mining I would be on a FIFO salary for life!
So we’ve got all of our data collected, and now we need to do something with it.
How effectively are you processing your data?
Do you have a clear system that’s mapped out to ensure that whatever processing techniques you’re using to get inferences from your data source, that they are clearly mapped.
Are they clearly defined, do the processes that you have put in place give a true reflection and a meaningful reflection of what’s going on.
Is it well documented?
Is the timing of your data processing going to help the team you are targeting to help?
Is it really going to inform them in the decision making process?
Timing of data presentation
Data Processing is a bit of a complex beast, but in reality, data processing is what enables people and our operational teams to make effective decisions.
If they don’t have a clear understanding of what it is that they need to do at the time that they need to make that decision, your data may as well just be put in the bin.
But it also goes the other way as well! Sometimes you don’t really need to have a consistent flow of data especially when it has to do with human performance.
People need time to work on one thing at a time, embed it and then move onto the next thing. Making each change a new habit it the way they work. This takes time!
Just know that the timing of data is important it might be good for retrospective views, that how do you meaningfully meaningfully change the perspective and the approach that people are taking at a specific point in time?
Formatting Data
The format of your data. How you lay your data out and set up your outputs.
This maybe should have actually gone a little bit earlier in this article, but you need to ensure that the format of your data is going to be sustainable for the long term.
Even when you think that you have nailed it I am sure that you will have to make a change.
Please keep this in mind as you go about setting your data up.
When you add other sensors and other data inputs into your warehousing solution, it is critical that you get this piece right.
If you mess this up, you’re going to create a lot of work for yourselves and your team down the track.
It is that simple!!
Data Visualisation
Some of us are amazing when it comes to understanding the numbers and being able to really quickly and easily identify exactly what might be happening in a data set.
But others we just aren’t that way inclined.
It needs to be spelled out for us.
We almost need to be hit by a sledge hammer by the data that enables us to go seeking further information.
Do not leave anything up to interpretation. Make your data visualisation super clear.
Super concise,
Your goal is to enable your mines operational team to get well set up in their approach to the delivery of whatever it is that they’re having to do for your mine site.
Supporting performance through data
It really comes down to enabling a frontline leader or operator or anyone else for that matter to very quickly and easily look at a data set or dashboard and know exactly what they need to go and do.
You don’t want to leave them sitting there trying to work stuff out.
Because they’re just not going to do it.
They’re not going to come back to use the dashboard.
They’re not going to come back to use your tool.
They’re going to move on and go and do stuff that they’ve always done.
And this comes down to I’m sure plenty of you have heard of it.
But PDCA
Please
Don’t
Change
Anything
More often than not.
This is why people get into that sort of headspace.
Data will never tell the entire story
One thing to remember is that the data will never tell the entire story.
But our job as leaders, in any mining business is to ensure that we understand what portions of the story are being missed.
How do we better go about telling that section or that part of the story?
Because if we aren’t doing that effectively and we aren’t telling the full picture, how can we really know that we’re improving where it really matters?
Focus on metrics that matter.
I love a good value driver tree and the reason why I love a value driver tree so much is when it’s effectively mapped out, you can clearly identify the levers that you need to go after that are going to deliver the largest results.
Focus on Mining Metrics That Matter
But when you’re focusing on metrics that matter, make sure those metrics matter to the person that is focused on them.
Because if they don’t, and they can’t influence them and they can’t impact upon them, then they’re not going to do anything about it.
Let me repeat that.
If somebody cannot impact upon a metric, they are not going to do anything about it!!
My last couple of points are around things maybe to consider outside of what we’ve spoken about.
Your mining Data Truth Set… It never lies
Firstly, live by a truth set.
Make sure that you have a truth set that’s been well tested, and well understood.
So is that when you make a change to any of your metrics, or any of the processes that you use to process data, that you can test the changes that you’ve made against that truth set.
Your truth set enables you to understand what is happening in your data world.
And my final point,
Automate your Mine Sites Data
Don’t let humans get involved when a well written piece of code can do it.
The code will be more accurate,
More reliable
And generally last longer!
Please, please please make sure that you automate your data collection, testing, processing and reports!
If this article has helped your or your team in anyway please share it with people who may appreciate it, or those that might like to have a different view!!