Hi, good morning, good afternoon. Thank you for joining us. I am Hernan Cianfagna from Crate.io, the company that developed CrateDB, which is an open source, distributed, database system, designed to manage large amounts of data and analyze them in real-time, and today I am here with Michael and Nixon from SPGo!.
Hi Michael, hi Nixon, how are you? Hi, how are you? Good morning. Hi, good morning Hernan, how are you? Warm greetings to Crate.io, thank you for organizing this webinar! Truly a pleasure to speak with you today. For those who do not know SPGo! they are a Chilean company, part of Petromin and they have developed a system called Live Conveyor for the predictive maintenance of conveyor belts in the mining industry, and today they will tell us about their experience, the motivations for building this, and the challenges they faced during the development. In case any questions come to mind during the talk, please write them using the dedicated Zoom Q&A feature and we will answer them all at the end of the event. To start, Michael, maybe you could tell us a bit about the background, the context we are speaking about? Well, first of all, I want to say hello to everyone in the audience, thank you for joining us today in this talk, which will be quite technical.
We will talk about how we implemented this innovation in South America and tell a bit about what has been our experience from the beginning, how the challenge started for us, and then a bit more about how we developed this idea over time and materialized it. First, we need to consider that as a mining country in Chile, we have a survey. The last one, from November 2022, has been carried out by COCHILCO, which is the Chilean Copper Commission. They survey mining companies worldwide that extract different minerals and they look in particular at who are the main buyers of copper. Considering
that information, we worked with this official data, and a couple of years ago, when we were working on some projects in the north of Chile where the weather is quite dry, very dry, even desertic, we realized that in some open-cut mines or open-pit mines as we call it, these are mines where the extraction happens by digging directly in the ground and this large pit is built bit by bit, this is convenient for some mines because it enables the use of very large machines to extract large volumes of raw minerals; In underground mines, the machines used are not that large, they are more medium-sized, but they do not allow to extract the volumes of copper that can be extracted with open-pit mining. We then realized that both kinds of mining require the transport of large quantities of raw ore. We then had a discussion with our holding company, Petromin which is a company that has around 25 years of experience in the mining industry. It is the sole dealer, let's say in Chile, for the services around fueling optimization for CAEX trucks. These trucks arrive to carry 350 to 430 tons of minerals so there are fleets made up of 50 going up to 120 trucks. So the amount of materials that
they can move in the entire fleet is of huge proportions. Well, when we face an underground mine, this kind of machines are not used and we realized that the conveyor belts become one of the main assets, let's say for both types of mines. There are technical inspections of the conveyor belts to monitor if they are working correctly, this includes looking at the status of the pulleys, their temperature, and making sure they are structurally sound and do not collapse. There are also checks to avoid fires, and maybe we can now watch some videos. Hernan, if you can play them, these are videos of some of the latest events that we came across about fires in conveyor belts. Yes, we got this video.
Could you tell us a bit about what is going here? Sure. What happens here is that these conveyor belts, while they were working, and it could be because of the rollers getting jammed, these start to overheat to high temperatures, very high temperatures, and this ends up resulting on ignition. Because of the materials at play here, we may be talking something like rubber, and these are materials that once they catch fire, it becomes difficult to extinguish it. Then we have fire incidents like this. There are additional complexities with some of these pulleys. They may for instance be installed at a certain height from the ground, and this makes it complex to check what is their level of vibration, which has some times led to the collapse of the structure of some conveyor belts over the years.
We need to consider that when a conveyor belt stops working, all work in the mine stops, the production stops. We soon realized that this was actually quite a complex challenge, not for a single mine, but for the entire region. Now our challenge is to be able to reach those conveyor belts that carry materials from the mines to the ships for our buyers, which can be for instance in China, Asia, and so on. To avoid this kind of issues, what is done at the moment are checks carried out manually, right? Sure, how the inspections are carried out at the moment? Well, when we were working in the north, we realized that the people that work on the inspections walk along the belts with temperature guns, and they go up and down these long conveyor belts. If we talk about a conveyor belt that is 500 meters long, there is no problem, but nowadays these conveyor belts reach long distances, some can easily be up to 12km long.
So walking the 12km distance. In a mining context, which let's say, like in a building environment, this is a complex environment because of dust, and a whole set of associated risks like, moving machines. For safety reasons, the people that work on this have to wear a large number of protection equipment, like a fire-proof jacket, safety shoes, special gloves, dust masks, and in case of underground mines, they use special masks, which you may have seen before. They are known as full face masks, protecting the whole face.
When the inspectors go inside an underground section, typically this will not be a flat horizontal tunnel, but it may be uphill and the effort required from the person doing the inspection is considerable. So, of course, walking 5 km, with all this safety equipment, plus the temperature gun, plus a tablet to register measurements, it is challenging. Then, to put this data in a database and manage to analyze it for a 12 kilometers-long conveyor belt, or even one that is 6 kilometers-long, and making decisions on this data, becomes quickly difficult. It ends up
not being preventive at all. It is not real-time anymore, we end up working with what is essentially historical data about the operations, not what is going on right now, and that is why we decided to explore this technology, to find a way to enable the mining industry to make decisions based on real-time data. This is what is going on right now in the conveyor belts, and we want people to be able to check this from mobile devices, from PCs, from a control room, all in real-time, 24 hours a day, mitigating risks. And fundamentally we are adding something very important for large producers, which is what everyone knows is ideal, which is to do maintenance in a preventative fashion, not just corrective, once something has already failed, but doing it to prevent failures from occurring. When you can foresee the issue because you understand the condition about what is happening with the conveyor belt, you can plan maintenance work.
Then if you know how each device behaves during the working day, you can with the predictive system we already implemented see how many working hours away is a failure on a component expected to occur. This is what this innovation is about. And here we have it, on this slide, we can see the system you have developed. Could you tell us a bit about the different components we see on screen? Sure, this is of course a simplified schema of a conveyor belt which we see here on top of these pulleys, or rollers, and what is labelled with number 1 is a schema of our sensor, which is installed on the support of the stands of the pulley. And here we also see our temperature monitoring on the shafts.
And what do we achieve with this? Imagine we got here a conveyor belt with a length of about 7 km, and there is a station every meter, and a separate set of rollers for the return of the belt every 3 meters more or less. We would be talking about something like 19,000 rollers in the conveyor belt. And if we want to check the temperature we need to install sensors on both sides of the belt, so the number of sensors is doubled, we are now talking about 38,000 sensors that need 24 hours monitoring for a full conveyor belt. This is like if we had a high performance athlete. They would be running all day but you would know their performance when they need to drink some water and so on. Basically, we want to preempt any sickness. Give them vitamins for instance at the right time.
A great thing about this system is that when it comes to installing the sensors, it is not invasive, meaning there is no wiring to do. We use wireless networking, an internal network between our devices and this links to our platform where we do the data analysis. And people may ask if it does not need wiring for electricity or networking, how long you think we need for this setup? It is easy, 2 to 3 minutes per sensor. To do an installation with wires, with fiber, would be complex. It could take months, even years, to deploy something this size, but with our approach, in 2 to 4 months we can have the system up and running with our platform, and it gets its power from solar panels. Very interesting of course, because, with this context, remote areas, desert climate, lack of connectivity, availability of electric power, and all that, I imagine that it must have been a challenge.
Nixon, perhaps you could tell us about this? I imagine that when you started with this project it was not obvious that solar panels and a wireless network were the way to go. Could you tell us a bit about the options you considered and what did you observed? Yes, I tell you Hernan, one of the main challenges we had, was one that you just mentioned, access to systems like power as well as internet. Regarding power, we designed the sensors using a system which includes a battery which lasts between 2 and 3 years, and these batteries can be replaced on each sensor. In order to power the wireless network, we installed solar panels, which are designed to last 20 years. This way we designed the infrastructure to bring the data to a data center where we store all the information.
We decided to go for this option instead of sending all the data to the cloud. We have all the data on an on-premises data center. This was for a number of reasons, mainly costs, but also internet connectivity. I can tell you that in these scenarios, connecting to the internet is very difficult, having an internet connectivity which is redundant and reliable is difficult. This is one of the reasons why we implemented this with CrateDB, to be able to have a database cluster running locally. We wanted to be able to store all the data, and be able to work with it locally, that is all the big data that the conveyor belt generates.
I understand, because as we were saying, there is a huge volume of data coming from these conveyor belts which can be several kilometers long, with sensors on each roller. How many sensors are we talking about? We are talking in average, about 38,000 sensors and these send a reading every 10 seconds. Approximately, in a day you can have an average of 228,000 records and this is just for one conveyor belt. For a month, we can talk about 9 billion records.
And to store this data over time for a year gives us about 120 billion records. So the challenge is not just about the amount of data to store, but it is also about the speed that we need for queries, because every 10 seconds we need to evaluate the condition of every sensor in the conveyor belt. For this, we needed our database system, our cluster, to be fast and robust enough to deliver this kind of information. In this case, we discarded other alternatives like cloud services and MongoDB, and it was CrateDB the one we decided to use. It has saved us, and it became an integral part of the solution we implemented. Happy to hear that. It is very interesting
what you mention. So, you considered cloud-based options, but unreliable internet connectivity made you discard them. You mentioned MongoDB, did you try any other systems besides MongoDB and CrateDB? We tried something like 3 or 4 database systems really before settling for CrateDB. We tried many database systems that were supposed to be geared to work with IIoT, but the one that worked for us, the one that came up on top, especially on data ingestion, and giving back query results with a relatively astonishing speed. That one
is CrateDB. Fantastic. So, you spoke about the volume of data at play, you mentioned before that it was important that queries could run quickly and, well, once you got the data into the database, I imagine the next challenge must have been building an alert system on top of it, right? Sure, so we had to consider everything around the speed at which information arrives and the querying, how we could generate the queries for the database.
Now, about the volume of data, this is the most astounding thing, CrateDB scales in a horizontal fashion, so we can for instance have 3 servers working and if we need to manage an additional conveyor belt, some additional projects, we can add more servers and the additional load does not impact the performance of the database or the performance of the system. Yes, this is indeed something we highlight as a benefit of CrateDB, the fact that it can be scaled horizontally. Hernan, in the next slide, the systems automatically runs a query every minute and displays in the top of the screen, the highest temperatures observed in average for each sector, the highest ones, and in the lower part of the screen it shows all the pulleys and all sensors on both sides of the pulleys. It displays a graph with all the thermographic information, indicating the different values with a gradient of colors, these align to the temperature intervals that come from the sensors. Excellent, this gives you full visibility of the full conveyor belt in real-time. Exactly. Unlike what happened with the manual checks
where in the best case you could take one measurement a day from each point, I suppose, right? And there is another thing, it constantly evaluates, if you look at this slide, it checks every 10 seconds, the temperature, the humidity, and the vibration. You can trace the data for the temperature for a month or a year, that is how the pulleys have been behaving over time, and you can see this for a specific section, you can see how the conveyor belt behaves, and what pulley system works best, what company brings you the best service, the best product. Here you have so much information available that you can dig into every small detail of the conveyor belt.
Yes, indeed. I see that we also have a geolocation thing here. I see that the customer can see a map of the conveyor belt and get the details of each segment of the belt. Exactly, this was added using directly GPS geolocation data that each sensor sends so the mapping is instant. Perfect, and yes this is another feature that we have, for the benefit of the audience, this is another feature in CrateDB, the ability to store geolocation data in the database and run queries around it.
Very, very interesting system indeed. I think this may be the leading solution on this front in LATAM at this time. We have done some research and we are the first in LATAM to implement an IIoT solution this robust for the mining sector. Also, Hernan, let's consider that we participated in 2022, in the Exponor trade fair in Antofagasta, in the north of Chile. We presented our technology, and of course the fact that we could meet the customers and present to them how a large conveyor belt, 12 kilometers-long, could be monitored with geolocation, being able to see exactly where the problems are, the fact that you could coordinate your logistics to carry out maintenance, and also being able to evaluate which sellers can nowadays give the pulleys system with the longest life expectancy, with the best uptime. That, at the end of the day, means less downtime, and more productivity.
And this is what we enable the customers to do. We realized there in the north of Chile how impactful the geolocation feature was. We realized that it is a very strong point. We have also been at the PERUMIN mining convention, also last year. There we spoke with mining customers, and we are at the point where it is imminent that we will be testing at South American level.
And finding that customers from countries such as Brazil, Finland, Australia, and the USA, could find here, in South America, this technology, and they quite liked it, was quite good news for us, and it left the impression on us that we are at the forefront of innovation in this sector. Yes, I can imagine it must be quite a satisfaction to receive this kind of positive feedback from customers. I was just about to ask you what their thoughts were, what you had heard from the mining companies about the solution.
So I am sure you will be very successful with this. Very interesting indeed. Hernan, another thing that is relevant to mention for us, is that in the last 3 years, since the idea was born, until now that it is available in the market, one thing that was complex was to get the certification for our product and sensors which is a compliance requirement, and here in Chile we already have the certification to use this.
We managed, let's say, to reflect that what we are selling is really, from how it is produced to its use, something that can be relied on, a good product, that comes with a warranty for the sensors covering them for 2 years of autonomy for the battery power. So I am quite happy with this innovation, and having a product that can be deployed without taking resources away from the mining company. Excellent. Indeed, it is always good to have some industry certifications to confirm the quality of the product. And let's now see if we got any questions from the audience.
Yes, I can see that we got questions from the audience. I am not sure how they appear on your screen, so I will read them in English for our international audience and in Spanish and I will also translate to the other language what Michael and Nixon reply, so everyone is able to get the answers. Here is the first question, I do not know if it would be one for Michael or for Nixon. Franco says: "Good morning, good afternoon, thank you for presenting the use case, considering the dust conditions in the environment, could it have been better to use wind mills? I ask because of the need to clean the solar panels." I will repeat this in English. Sorry about that. I will read all questions in English and in Spanish and I will translate back the answers so that everyone has an opportunity to get all the information.
So, we were starting with this first question, where Franco asks if because of the dust conditions in the environment, if using windmills would have been better than using solar panels to avoid having to worry about cleaning the solar panels. Ok, maybe Michael do you want to answer this question? or perhaps Nixon? Sure, Hernan. Hello everyone. Thank you for attending this webinar. I will tell you that when we pose ourselves the question of what was the problem that we wanted to solve for our customers, we always kept in mind that it had to be something very easy to maintain, very simple to operate. We knew that if the product would have required complex maintenance that would have translated into high staff costs, because of the skills and high professional standards required.
So we wanted to have something very simple, for instance regarding the maintenance of the solar panels, a very simple cleaning process. So we can reduce the number of staff needed for this, but also lower the costs in terms of not having to have highly skilled staff to maintain a complex system. So, basically the answer from Michael is that they considered alternatives, but they were very mindful of the maintenance costs considering all dimensions, what it would mean in terms of the qualifications the maintenance staff would need to have to look after these devices. Considering all this, they decided that the solar panels, even though they needed some cleaning, that could be done very easily, very quickly and it did not require special qualifications from the staff, so that is why they decided to go for solar panels. We got another question here. The next question is: "What framework is the front-end based on?" So the next question, which I just read in English, is from Roland. This is probably one for
Nixon. He asks about the UI, the front-end. The question is: "what framework was used?" It was developed with Node.js. It is based on Node.js and HTML5. It is a web application.
Thank you Nixon. Nixon says that it is actually a web interface and it is developed with Node.js. We have another question, from Stefanie and she asks if there are any plans to market this system for other industries, such as refineries or logistics? So in Spanish, the question is for Michael and it is if you have plans to market the system for other industries besides mining? Refineries and logistics for instance.
Well, our experience is really very focused on mining, refining, and construction. Today mining is the main industry for us for this solution because of the need being there, but refining is of course interesting because of the way the industry works, connection to ports, and the logistics involved, so similar needs and lots of potential like mining. Then yes, refining, it is in the direction we are going, we will get there. The idea is to put sensors on all moving machines so that the client is empowered to make decisions. So, yes, they are considering that, especially refineries because there is the connection with the fact that they are usually based near ports, and also the solution could apply to any moving machines in general.
The idea of the sensors that they have developed, and the system, could apply to these kind of environments. The next question is from Steve and he says: "Thank you for sharing the very interesting use case. You mentioned that you looked at MongoDB. What other databases did you look at?". So this would be a question for Nixon. It links with a point
we discussed during the chat a few minutes ago. Steve says that you Nixon mentioned you considered MongoDB and he asks specifically which other systems you considered. We considered InfluxDB, Cassandra, and also Firebase. Thank you. Nixon says InfluxDB, Cassandra, and What is the name again of the other system Nixon? Firebase.
We got another question from an anonymous attendee. This one is in Spanish so I will translate it to English. They say: "What is the reach distance of the wireless sensor? What protocol is used for communication?". This question is about the wireless sensors and asks what distance it can reach for transmission and what communication protocol it uses. I see, the sensor can transmit up to a distance of 150m. The technology is Bluetooth Low Energy and it works with the MQTT protocol.
So, 150 metres is the reach and it uses a low energy Bluetooth and MQTT. We got one more anonymous question in Spanish. Another question, this time in Spanish, they ask if they considered the changes in in ambient temperature to decide the thresholds for the alarms? So this other question asks if you consider changes in ambient temperature to define the alarm thresholds for temperature readings. Sure, every sensor takes a temperature reading of the object and in addition to that it also takes a reading of the ambient temperature. So we have 2 temperature readings, the one of the environment, and the one of the object itself. Excellent. Hernan, one more
thing we consider in our product is the altitude. In Chile, we have mining sites at many different altitudes. So, of course, in a site which is at 4200 meters over sea level temperature is effectively lower, but being exposed to sunshine and atmospheric pressure, the apparent temperature is much higher. Sunlight radiation feels much stronger, so we did tests, when choosing the sensors, to make sure we could make a distinction, based on the site's altitude to avoid false positives, to not trigger alarms by mistake.
Perfect. So, first, Nixon explained that they are actually taking two temperature measurements at each point. So they are sensing the ambient temperature and the temperature at the point in the machine that is being monitored. Then Michael elaborated on that and he said that they were very mindful, especially because in Chile there are places situated at different altitudes, and the perception of temperature can change dramatically depending on where you are.
So they tried different sensors and they were very mindful of this point because they wanted to make sure that they could avoid false positives. There is another one in Spanish again. This one asks how the sensors are attached, if it is with magnets, with screws, with silicone, or something else? This question is in Spanish, and it asks how the sensors are attached considering the temperature? Is it with magnets, with screws, with silicone? Something else? We use 2 attachment processes, the first involves a silicone that can withstand high temperatures, and we use this silicone to attach the sensor to the frame. So, the body of the sensor is attached with silicone, and the the contact sensing area is attached with special clamps that allow to cover shafts. And then, the second process I was talking about, we use clamps to keep the body of the sensor in place until the silicone has dried 100%, which takes about 3 hours after installation.
This is to avoid having to stop the conveyor belt. The fact it gets attached with clamps allows it to work right away. Perfect. So I will give a short translation.
So, they use a special silicone that can withstand high temperatures and they also attach it at the same time with clamps. So, let me explain. The silicone takes a couple of hours to actually harden so that is why they use this second mechanical assistance, so that the sensor can start working right away without having to stop the conveyor belt. And the next question is in Spanish, from Ans: "What type of sensors they utilize, what type of communication they use, that was low energy Bluetooth, and how they are installed?". That is what has just been answered. This other question is in Spanish and it is similar to the one before, it asks what kind of sensors are used. That part of the question is new. Then it also asks what protocol it uses for communications and how it is installed on the frames, which is what Michael just explained.
Regarding the type of sensors, they are sensors for IIoT which work with IoT. Perfect. So, it is specific IIoT sensors. The thing is that we designed the sensor ourselves. We designed it to collect the readings we required.
So, it is not an off the shelf sensor? You built it yourselves? Yes, we designed it specifically so that it could collect these readings. Great, I understand. So, it is not an off the shelf sensor. It is something that they have built, they choose the components, but they built it specifically for this use case with all the considerations that they explained earlier.
And then we got another question, this time in English. I will read it because I think you cannot see it. The question is: "If being able to write the queries in the SQL language in CrateDB, if that in retrospect was useful, or if you could have worked with another language?". So this question would translate in Spanish as whether being able to write queries with SQL was useful, or if in retrospect you think that you could have written them in another language. I think that for us whether writing queries was difficult or not was not really a concern.
The main thing for us was the scalability of the platform. The scalability of the database. I mean, writing a query in SQL is easy. The challenge is when we have millions of records and need to run a query over millions of records in milliseconds. Perfect. So Nixon said that yes, writing queries in the SQL language is easy,
but for them the main reason to go for CrateDB was actually performance, that they could get answers in milliseconds. That was the main reason. Steve says thank you for sharing. And we have one more question. I think it is the last one. So we got a last question from Vasyl. They ask: "What brands of sensors were used?". They also asked:
"Are they ATEX compliant? What was Ex zone in the places where sensors were installed?". Not sure what he means by that. So we have a last question, this one is in English. They ask again the brand of the sensors, and I understand there is no particular brand as they are custom made, but then they also ask if the sensors are ATEX compliant. That is an acronym I had not heard before. And then they ask
where the sensors have been deployed? Well, I will start first with the compliance question. To be able to put the product in the market, for instance in Europe, we got of course European regulations. In the US they also got their regulations. And so on for each
country, like Chile got its own internal rules. Usually there are rules to certify a product for use. So it will depend a bit on which country would need our solution for us then to pass a 3rd certification process as we already have 2. We would not have any problem with that if we have demand from abroad from a country where ATEX compliance is required. Now regarding the type and brand of the sensor, like Nixon explained, there is not really a brand as we built the sensor to our requirements, the requirements of our customer, so regarding our platform and technology, finding something similar, it will be difficult. I do not say
it is impossible to replicate, but consider we have been working on this pioneering technology for about 3 years, facing many challenges, we learnt a lot along the process, there was intuition involved, but every problem we faced took time and work to figure out. So I do not know if you could find something like this. And I believe that you could look if a similar technology exists, but you need to consider which system gives the best results for your company, and which technology gives you up-to-date data, real-time data, which one gives the best availability for your conveyor belt, which one increases your productivity and reduces the number of people that you need on site which at the end of the day reduces the risks, the risks of having people exposed to moving machines. Excellent. So Michael explained that, of course, each country has
different compliance requirements. At the moment, they got 2 certifications. They do not have ATEX in particular, but if a customer requires a particular compliance certification, they are happy to apply and go through the process to obtain additional certifications.
They were very careful when developing this solution. They always had in mind that they wanted to develop this to improve the safety of the inspectors, of the people working on the mine, and the productivity. They are confident they could pass another certification process if required, so that is the answer to that question. We got another question, that came in after I said that the previous one was the last one. I think this could be translated in different ways into English, so perhaps I will correct myself after we get the answer from Michael and Nixon. The question seems to be: "What IP protection rating does the sensor have?". It is a question about
the sensor and how it looks physically speaking. Let me ask Michael and Nixon and then I will translate back. The question is in Spanish, I am not very sure what it is about exactly, but Ans asks: "What level of Ingress Protection the sensor has, how it looks physically speaking and if it is a "transmitting sensor", an integrated one, or a remote one?". Well, it is certified for dust, acid fog, water, and also to withstand being hit. That about the
durability of the sensor structure. Regarding data transmission, I am not very sure what the question was about. Thank you Nixon. So Nixon explained that the sensor is certified to withstand exposure to dust, also acid fog, being hit to a certain level, so yes they build it to withstand the harsh conditions there in the mines. About the second part of the question,
he was not very sure what the question was about. So if you want to elaborate further in the Q&A section we can support that. In the meanwhile, Franco says: thank you, very interesting use case. We received a message from a person saying thank you for sharing. It seems we do not have any more questions at this time. So, we will share a link with you
to download a white paper about IIoT technologies which you can find interesting. If any question comes to mind, you can reach out to us with the contact details you see on the screen. Please do not hesitate to contact us by e-mail. If you want
more information about CrateDB, I invite you to visit crate.io website and for more information about SPGo! and the Live Conveyor system you can visit spgo.cl. One last thing, if you want to try CrateDB we have the CrateDB Cloud service, which is a service offering CrateDB in the cloud, fully-managed by us, so you do not need to worry about the infrastructure or about monitoring. Just by registering on our website you can automatically receive a credit of 200 USD that you can use to try it out. I hope you found this webinar interesting. Thank you very much Michael, Nixon. Thanks a lot SPGo!
for having shared their experience. And also thank you very much everyone for having joined us today. Have a nice day. Bye.
Thank you everyone. See you Hernan.
2023-02-07