218 - The Geospatial Planning Episode - The EV Musings Podcast

Gary C:

Hi, I'm Gary and this is Episode 218 of EV Musings, a podcast about renewables, electric vehicles, and things that are interesting to electric vehicle owners. On the show today we'll be looking at why charges get installed where they do

This season of the podcast is sponsored by Zapmap the free to download app that helps EV drivers search plan and pay for their charging. Before we start, I want you to thank everyone who's tuned in and listen to season. This is the last quote unquote topic episode of the season ie one where I talk about a specific subject. The remaining two episodes are focused slightly differently, and they include the season ending round table episode. So thanks for listening in, retweeting, and helping to promote the show, Much appreciated.

Our main topic of discussion today is locating charges. Now if you've ever been on social media, and who hasn't nowadays, you'll see that the charge point operators regularly announced the openings of their latest charger locations. If you follow MusingsEV on Twitter or X, Twitter X, whatever it's called now I tend to retweet these things for a slightly wider audience and almost invariably, there's at least one person on every reply that posts a variation of 'Why won't you come and install charges in Lincolnshire / Cumbria/ mid Wales/ insert your location of choice?' Now I heard a stat last year which said something along the lines of one in three charges in the UK is located in London ie within the M25, and that's great for London, but it's not so great for places like Bedgelert in Wales or Binbrook in Lincolnshire. Dan McLaren from bp Pulse said on this very podcast that 75% of the UK population are within five miles of a bp Pulse Charge Point. Again, that's great for those 75%. Although I suspect a lot of them are in London. But what about the remaining 25% Or as we like to call it in plain figures, 15 million people? Now we've talked with many charge point operators on the show over the years, and I've talked to them all about how they decide where to put their chargers. The answers are multiple and varied. But they all boil down to one short sentence. We can't put them everywhere immediately, so we have to put them where it makes most sense. Now within that sentence, there are a number of other questions that present themselves. Who decide what makes most sense? What are the drivers behind this? And is finance an issue? Now a lot of these discussions take place behind closed doors and depend on a huge number of factors to determine charger location. And despite what you think in some cases, charge point operators don't just throw a pin in a map and decide where these installs go. There are numerous factors that go into them. And one of these factors is what's called Geospatial planning. Now, geospatial planning is the practice of working out where you need to install charges. You have some commercial aspect to look at the physical demand or physical need, and then working out which location is going to emerge first, and what degree will at least ask my definition. Let's have someone who knows,

Charlie G:

Geospatial planning clearly has lots of applications in lots of different industries.

Gary C:

That's Charlie Gilbert.

Charlie G:

I'm Charlie Gilbert. I'm a partner at Field Dynamics

Gary C:

Anything else to say on geospatial planning? Charlie?

Charlie G:

Town and master planning is inherently geospatial. Transport modeling is the same, as well as retail planning. Where do we put the next Aldi as a supermarket? Where's the next Gregg's going to be? And I think the same is absolutely true of EV infrastructure. And there are many different facets of that, which I'm sure we'll step into. My mates in the pub tell me surely it's really easy, you just pick a busy road, and that's where all the traffic is. And at face value, that's partly true. But actually, when you get into the detail, it couldn't be more different. But fundamentally, it's mapping the relationships between supply and demand, and for our clients, and that really depends on their use case and what they're trying to achieve. It's very much helping them with that decision making process using location as an aspect,

Gary C:

Now let's talk to another player in the field. Geotab. Here's Oliver Holt.

Oliver H:

My name is Oliver Holt, Sales Manager for UK and Ireland for Geotab. I'm an interface between our customers partner ecosystem, and I tried to understand how best to deliver our solution to our end users.

Gary C:

And how does he define geospatial planning?

Oliver H:

Geospatial planning, I think is an interesting terminology. So the way that we view that is, it's bringing together a number of location-based technologies to understand the landscape or an area or a municipality, to then advise future development and optimize infrastructure planning in a certain area moving forward.

Gary C:

Now, Field Dynamics and Geotab are both players in the growing field concerned with capturing, mining, and selling data to help businesses make business decisions on something which is primarily non financial - although money obviously plays into this in some form. The key word in that last sentence is data, lots and lots of data. But Geotab, and Field Dynamics have multiple huge datasets that they can access to help them provide information for customers. So of course, lots of data, lots of analysis. How does AI plays into this? Here's Oliver H: again,

Oliver H:

Interesting question actually. It's a quickly evolving landscape, especially in regards to AI So Geotab probably takes a two pronged approach to this question. We do apply AI in certain areas of the data analysis, and especially on that sort of back end analysis of the raw data, AI has got a huge application, especially in what we're doing today. But we always start at the point of the data collection. So essentially, we're collecting large quantities of high fidelity data, which then gives us a starting point to make really informed data led decisions. Following that we then draw insight from that data, both sort of in the old traditional manual way of understanding reporting and things like this, but then, increasingly, so applying AI to help us draw that insight as well. And then be quite prescriptive with particular action points to take from the data to improve, you know, baselines, emission factors, and things like this.

Gary C:

Field dynamics do much the same thing I asked Charlie Gilbert about the datasets, where they come from, and whether AI plays into this.

Charlie G:

Yeah, I think that's a really good question. I think it very much depends on what clients are looking to achieve, the use case in question. We've always been a little bit nervous about the concept of AI. And we hear about it so much in the news, and in the media these days about the direction of travel, I think, absolutely. That's a key part. But ultimately, AI relies on lots and lots of data. And when we first started getting into looking at easy infrastructure and planning, yes, there was lots of location data. But there wasn't really any data about how the infrastructure and the assets are being used, what are their patterns of life, all of those sorts of things. So I think AI is absolutely beginning to become a real part that you need to have a training data set to model on top of to do those forecasts and predictions. So our view has very much been that we take a no more than one step away from the fact approach in our analysis. So because we like measuring things and analyzing things geographically, we take a very hyper-granular view when we're looking at the data. So to unpack that, and give you a couple of examples. And I'll just talk a bit about how we came up with one of the datasets that we work with. I went to have a test drive. A Tesla showroom. And the last question that the guy asked me in the showroom was' I take it you'll be able to park your vehicle on your driveway, sir?' And I said, Yeah, sure, no problem at all. And lo and behold, when I got back, there wasn't a space to park. And as somebody who's got a geography degree, 20 plus years in geospatial planning, I got it completely wrong. So I thought it'd be quite a good idea would be why don't we create a data set that looks at the possible space for every single household in the country, so all 27 million households? So that's very much what got us on this hyper granular journey are saying, Okay, well, in order to have accurate forecasts and models, and there's great merits of doing things up, bottom up and top down, having this really solid foundation of data has been really, really key to us. And we've now built out that dataset, which is called EVmap. We have about 150 local authorities who were using it DNOs, charge point operators in their planning. And it's now becoming a very much an industry mainstay in terms of Field Dynamics are the guys who look after that on street household data. So to pare it back, we have a whole suite of datasets, and we've got five or six core datasets that we're really working on very hard at the moment. But we've got a list of about 30 that we have in the pipeline to create and there's some weird and wonderful things in there as well.

Gary C:

So we've got huge datasets and people wanting to find some signal from within the noise. How do these companies interact with customers? ChargePoint operators for example, is it a case of we'll sell you the data and you can extract what you want?. Or do companies like Field Dynamics provide consultancy in these areas,

Charlie G:

It depends on both. So the skills that a Charge Point operator clients have vary. Some of the more established providers have their own data science practices, they have their own teams of analysts, they have their own tools that they built in house. And then the requirement for us then is to fill the gaps, plug those holes where they are, that are lacking the granularity or lacking the coverage. So that's the first piece, we also have clients coming to us and saying 'Right. Goal Seek. Where should we go?' Look at the whole of the UK and where are the best places to start to build out our footprint and expand our footprint? What's the competitive landscape? What are the power requirements? Where are where are the on street households? Where are the people who are doing high miles? Where are the fleets? All of those different constituant parts come from different elements and different datasets. And what we focus in on is a) bringing that data in so that we can make sense of it. One of the challenges that I think a lot of organizations have is there's so much data out there, trying to make sense of it and saying okay, so what? What does it really mean is really, really hard. And that's where our consulting practice delivers a lot of work for clients to help to answer a specific exam question. So typically, the brief is set by our customers. And it might be I have a portfolio of X locations, or I have a goal seek approach, or I've already got some kits in the ground, and I want to use some of the utilization data from those charge points to understand and then map the trends on to where areas are that would meet those similar characteristics to build out a particular market niche or focus area.

Gary C:

Oliver H: from Geotab says something very similar.

Oliver H:

So it can be very tailored to any end user, including charge point providers, depending on the information which is required. So in the European space, we've done a lot of work with local authorities, particularly in Madrid. So although I'm speaking to local authorities, as opposed to charge point providers, but I think the use case is probably fairly similar in that understanding the activity of vehicles, particularly fleet vehicles, within a certain vicinity, or in a geospatial area gives better insight into that planning element of where the infrastructure needs to go. And it can monitor just trips, activity levels, dwell times. And then what Geotab is quite strong at doing is overlaying multiple data points at the same time. And I think this will be interesting to Charge Point providers as well. So not only can you understand how a vehicle is being driven, how long they're being driven, where they're being driven, we can then overlay certain data points against that to then understand potentially fuel levels at the time of dwelling evey battery levels at the time of dwelling. So for vehicles dwelling in a particular space, and it's got a low percentage of charge at that time, potentially, that's an opportunity there where they could have done a charge if there was charging available. And I think that kind of insight would be quite interesting, particularly to charge point providers, as well as the local authorities, which we've worked with in Madrid

Gary C:

We'll come back to the data itself in a short while, but I wanted to know how this actually works in practice. So I had to ask the question that all my listeners will probably be screaming at their devices. Now. Why is it the charge point operators are not putting charges in places like Lincolnshire, mid Wales, Cumbria, or that stretch between Carlisle and the Scottish border? Is it because the planning is telling them not to? Or are you telling them to put them there and they're overriding that, for financial reasons? Here's Charlie Gilbert from Field Dynamics?

Charlie G:

oment, but we started that in: 2019

Gary C:

hey have telematics data from: 1000

Oliver H:

Absolutely I'd say from a vehicle side, specifically, Geotab's was very strong at this. So we can give you as much information as you probably require as a decision maker in infrastructure planning from a vehicle site. So where are the vehicles traveling? How long are they dwelling? Overlaid with that engine data, which we touched on earlier. I think like you said, there's other things to consider which might need, which is where all sorts of partnership and ecosystem come into play here. So you mentioned about maybe capacity on on a certain site for drawing energy from the grid. There's a lot of capacity challenges that we see across our customer base, which actually impacts a decision as to where they're going to place the infrastructure, whether it's viable site or not a viable site, depending on how many vehicles are operating there. So I'd say in this case, and direct answer to your question, Geotab can provide very strong vehicle side data. And we'd love to sort of partner and overlay that with someone who could advise on on the on the ground side of things, if that makes sense.

Gary C:

So let's take a specific example. What about, say, a supermarket chain with a limited number of sites, variable power availability, and a desire to install rapid charges *cough cough* Smart Charge? How would Field Dynamics approach something like that?

Charlie G:

So clearly, if we're looking at a supermarket, there are a lot of fill- physical elements to that. There's a lot of data that supermarkets have about who visits, what their client, what their type of typical footfall of their stores are. Some of those supermarkets have petrol forecourts associated and on their retail estate area. So what are we going to do about that? So there's this kind of formulation of, of an approach and a future state business model that we'd want to understand first, and then we know what we're shooting for. So here we're combining a range of different datasets, all the datasets that I talked about in terms of the demand, and looking at how that plays through. But really, this is now starting to play into analyzing existing usage data from those organizations if they've got either existing charge point provision, or using broader assumptions and, and model inputs to give a view to say what does good look like what is a typically busy EV charging forecourt or charging provision in a particular supermarket? And if I step back and look at the patterns of life, and take it back to the data for a minute, we travel about 500.. 5,500 miles a year, on average for every vehicle. That's what the MOT data says. So that's about 106 miles a week or something like that. So if we're looking at that number, then most modern EVs do a couple of 100 miles. So that really means that they need to charge up once every couple of weeks. So a supermarket is absolutely a brilliant location. But the charging infrastructure needs to fit in with that dwell time and help somebody is going to use that hub. And if there's lots of data about the supermarket, and what the sort of patterns are, and the opening times and access and all that sort of things, what is in the nearby vicinity, that is also going to increase in that- draw traffic to that area. And that's where we're playing in things like demographic data, were looking at where EVs currently have some level of penetration, there's obviously lots of caveats about the DFT data and registered keepers and registered owners. And we do lots of work to kind of work through that, but having an understanding of that side of things, but because we've got the MOT data and we understand that make and model aspect of what sort of vehicles are in the area, we can then jump across and go, Okay, well, what sort of vehicles are likely to be using that infrastructure? Is it going to be taxis? And if we look at the classic kind of taxi fleet, and there's some of the rod- ride hailing some fleet that's out there, those vehicles don't have the ability to take up Ultra rapid charge. So it might be a 150 charger, or 300, or something like that. But the vehicle can only pull 30. And I know this is something that we've we've talked about previously. So it's also providing that lens to say, 'Okay, right, what are the best? What is the best type of charging infrastructure to put in those sites?' And overlaying then the kind of forecast state, so overlaying things like ZEV mandate in and understanding what's that trajectory of the second hand car market?

Gary C:

Let's loop back to the data, shall we? Now I've worked in IT in the dim and distant past. And the one thing you learn pretty quickly when you're looking at computer systems and databases, is the old adage of 'Garbage in, garbage out'. With these two companies looking at capturing and using huge datasets from multiple different sources, how can they ensure good, clean, rationalized data, ie data with no multiples and no blanks where there shouldn't be blanks? Well, let's look at data sources for a start. here's Oliver H:.

Oliver H:

So I'd say it'll be interesting to unpack what the definition of telematics data is. Because again, I think just amongst the general population is the idea of just black box technology it's just GPS based, maybe some accelerometer in terms of driving style. And because there's a heavy association with insurance, what I'd like to do is expand what we mean by telematics data. I think this is where we can bring a lot more value to this sort of electrification and sustainability journey. So other things such as pulling fuel data from a vehicle can obviously really influence a sustainability agenda for a fleet, I'd say what we can also pull from a vehicle in in a lot of cases is the environmental situation at the time. So using the sensors on a vehicle, if it's possible that we can decode it, we could potentially pull the temperature, local temperature, street by street, city by city area by area, which again, could influence the planning for implementation of particularly EVs, because we know that EVs are impacted by sort of local environmental factors as well.

Gary C:

Now with this sort of data, there's not always a great deal of cleansing that's needed.

Oliver H:

Yeah, interesting. I'd say it's built in by design. So the way that we collect data from our devices themselves is via a patented algorithm, which is patented to Geotab its own technology. And it's called the Curve algorithm. Essentially, this is an algorithm which determines the point of data collection, the validity of that data and decides whether to keep and transmit or to discard at the time. And what this means is that we have a really efficient, but really accurate way of transferring data from the vehicle to our cloud environment. And obviously, it's in the cloud environment when we draw insights from the data. So when you say if there's a function within Geotab, to ensure data accuracy, it basically happens at the point of collection on the device itself is built into the firmware which we deploy to all devices.

Gary C:

For Charlie Gilbert at field dynamics, this is a whole different kettle of fish.

Charlie G:

Data quality. It's a big topic. It's a big topic, and we spend a lot of our time looking at it. It depends on the project. If we're working with some of our standard datasets, they've been through all the QA processes, they've been through that check. So when we do our EVMap dataset, we have students that run around and compare our dataset to the real world. We do Google Streetview surveys, as well to ensure that we're improving that accuracy and driving it. We also have a solid set of assumptions that clients know about how the data has been created and built through. But when it comes into organizations operational date, that's where we really have to focus on the DQ side, Field Dynamics has called Field Dynamics but actually we did a lot of work before we got into Net Zero looking at field operations, and trying to understand how to drive greater efficiencies, save utilities and clients and people who have large fleets money with that with their operations. And I would say we'd spend about 60% of a project, actually sorting out the data, taking data from telematics data, taking data from work order management systems, taking data from scheduling systems, bashing that all together, and playing that back to clients. So they go, okay, is this what your operation looks like, from what you know? And often, the operational leads in our clients, they know the data inside out, they know the sights, they know the locations. So what we always do is go, is this what you're seeing, and they go, yeah, that absolutely makes sense. And that's where we unpick things in a lot of the data that we work with, with clients. So systems wise, when we're looking at utilization data, you get all sorts of weird stuff, looking at utilization data, you get transactions that haven't happened, charges that have failed users that haven't been able to log on to the application as as as the kind of process should work. And it's about cutting all that out so that we're not then driving a slightly different or skewed result out of the back off. But part of the the iteration is going right. Okay, well, what does it look like? And yeah, okay, there's something weird going on here. Why have we got lots of usage in between one and two in the morning, or midnight, oh, Midnight's a default value. And it's those sorts of nuggets of information that we're able to pick up, and then we'd out so that we've got a really solid operational data set. So typically, that's from a charge point operator, it's data from their own CPMS. And that tends to be very systemized and structured. But from an operational perspective, sometimes we get free text. And we have to do text recognition to try and work out what type of jobs that was. There's all sorts of weird and wonderful techniques that we've used. But ultimately, when we're looking at things like fleet transmission is who went where, and did what, when, gives us a view of those patterns of life, the patterns of life of a particular vehicle in a fleet, whereas a Charge Point operator and their their back office system tends to be more structured. But there's different tables of levels of aggregation. There's lots of stitching that we have to get to look at, to build out that profile, and then overlay that with some of our own forecasts into rhetoric.

Gary C:

So, a couple of takeaways from this discussion. One, there's a huge amount of disparate data from multiple sources that go into deciding where the best place to locate a charter is. Some of this comes from the CPOs themselves. Some of its available to members of the public, often at a cost, and much of it is proprietary with companies like Geotab, using telematics data from fleets of vehicles, to build up patterns of movement, and understand what charging needs best match these patterns. Secondly, the sheer volume of data means often a company such as these will end up iterating their results, they can gather together a bunch of data, extract information from it, and present it to the client. And this data may be useful as it is. Or it may mean they have to go back a couple of more times at a different level of granularity to get the sort of results a client is looking for. And thirdly, what's also interesting is that this data from companies such as these, it's available to anyone who can afford the data and the associated consultancy. Charlie Gilbert told me that they're currently working with any number of charge point operators, although obviously couldn't name names for commercial reasons. But Field Dynamics are also working with councils and local authorities to help them understand the best places to put charges in their local areas. Geospatial planning and all sorts of work that companies like Geotab and Field Dynamics do it's a massive concept, it's a massive project or a set of massive projects. And it's not one that we can really do justice to in just one short episode. But I hope the discussions we've had today give you some idea of the sort of work that goes on behind the scenes when it comes to things such as the best places to locate charges in the country. We'll come back to this topic in an episode next season. But for now, I'd like to thank Charlie Gilbert and Oliver H: for their time. This season we're looking at raising the awareness of carbon literacy with our listeners Herman way we're doing this for the carbon fact, as read by carbon literacy trainer on Snelson.

Anne Snelson:

Ordering online and not in a hurry? Save emissions by getting delivery of all items at once. In fact, supermarket and online shopping delivery is usually better than going yourself. Research suggests it could produce three times less emissions because each man replaces several cars

Gary C:

big blocks of carbon to over: 4000

Charlie G:

We've always been a little bit nervous about the concept of AI and we hear about it so much in the news.

Gary C:

Thanks for listening. Bye

Share Episode

Shownotes

Transcripts

Follow

Links

Chapters

Video

More from YouTube