00:01I'm one of the government employees that Ray was making fun of earlier, the slow government employee.

00:08First off, good afternoon, everyone. I'd like to thank, before I start, Peter and the planning committee...

00:13...for giving me the invitation and opportunity to talk, specifically about this project.

00:17It's one that we're all quite excited about, and I think this is the perfect venue to speak about it.

00:24It's a project that we've been working on in our province for about the last year with Remsoft, both of our home provinces.

00:30It's a big one, so I'll step right into it.

00:34I'm going to start with just talking quickly about New Brunswick.

00:36So New Brunswick is in Atlantic Canada; it's on the other side of this continent.

00:41About seven million hectares, 17 million acres, one-sixth the size of California.

00:47Seven hundred and fifty thousand people pretty well spread out around the province, mostly in rural areas.

00:53And we're 85 percent forested, the highest proportion of forestland in any province in Canada, so the forest is important to us.

01:00So I could go on about New Brunswick all day; I'll stick to things that I think people here are interested in.

01:05One thing that might be of interest is my department, the Department of Natural Resources...

01:09... and Forest Management Branch that I work in, was actually the first licensee of ArcInfo in the world.

01:16We were the first licensee, commercial licensee.

01:19In the early '80s, 1982, we were just coming out of a spruce budworm epidemic...

01:23...and through lots of analysis, it became clear that in order to sustain the harvest levels...

01:28...that we were harvesting at the time, we needed to plan our harvest spatially.

01:33And at the time we were collecting a new inventory, and it just made sense to move toward GIS...

01:39...to house that inventory, and we acquired ArcInfo.

01:43As a result, we were the first jurisdiction in the world to produce a fully digital...

01:47...geographically referenced forest inventory of the entire forest.

01:52So of course, talking to the guys that were around at the time, I heard all kinds of neat things this week about...

01:57...it was, you know 300-megabyte hard drives the size of dishwashers...

02:01...and quarter-megabyte boards the size of a checkerboard.

02:05So there was some neat discussions this week when I was asking them about it.

02:08But I think it's testament to the fact New Brunswick's always been really interested in using technology...

02:14...to better understand the forest, understand our resource...

02:17...and to try to get the most value or extract the most value from the resource.

02:22And that's really what the project is about I'm talking about today.

02:26So our management planning process started back in the early '80s, as I said, and it really hasn't changed that much.

02:32The process hasn't changed. Information has changed; we're getting better and better information all the time...

02:36...but it's probably like most places, we conduct aerial photography on an annual basis.

02:42We collect ground information to improve that photography.

02:47We load our growth and our yield into forest estate models.

02:51We use those models to block the forest, and then we go out and harvest it.

02:55And, you know, things really haven't changed. For a long time, we've used Esri products to house and process that inventory.

03:02We used Remsoft solutions to calculate our AAC and to go block the forest.

03:10In recent times, though, we've found a couple of chinks in our armor.

03:14We've, you know, given the past few years and the situation in the forest industry...

03:19...the last piece of this process, we're missing some tools.

03:24Specifically, we've found three shortcomings. One is that we do a poor job in predicting the product supply out in the forest.

03:32Really good at volume; we can tell you our confidence levels for any hectare in terms of total volume by species...

03:38...but do a poor job at products.

03:40And that's ultimately what our mills consume is products, roundwood products, stud wood versus sawmill, or sawlogs.

03:47What we don't do a good job at is understanding the consumption requirements of the mills.

03:52So what's the top diameter and length of this mill's diet, and how much do they need to run profitably?

04:00And we don't spend a lot of time with location in terms of supply or demand.

04:04We don't make that connection between where the supply of a product is and where the mill that consumes that product is.

04:11So that, we think, if we can nail down those things...

04:15...it'll take us far in understanding the economics of the forest sector in our province.

04:21And that's really what this project attempted to do is really, really try to fix those chinks in the armor...

04:26...to address those three things, and it's no coincidence that Esri and the Remsoft solutions were a big part of that.

04:36So to describe in a little bit more detail the project, I'm going to start with describing a little bit about the forest...

04:42...and forest sector in the province, just sort of quickly, but that sets up the challenge that was presented to us...

04:49...a sort of problem statement for which this project was designed.

04:53Then I'm going to get right into sort of talking about the solution in the body of the presentation, the model that was developed...

04:58...give a quick policy example about, you know, how this model might be used to address policy solutions...

05:06...and then how we're going to move forward with it.

05:09So I want to describe the province's forest and forest sector from a perspective that I think...

05:16...sets up the requirement for this project, and that's complexity.

05:21And I'm not trying to set New Brunswick aside - you know, forestry's complex in general; I think it's complex everywhere...

05:27...but there's a certain set of conditions that makes New Brunswick, or makes this project particularly challenging and rewarding.

05:37So first is ownership; our ownership is complex.

05:39We have 50 percent of our forest owned by the public, by Crown; it's managed by the government.

05:46It's split up into six licenses there in green, the six different shades of green.

05:50So six different licensees with different management objectives.

05:54There's seven industrial freehold landowners, so those landowners own large wood processing facilities.

06:01And then 30 percent of the forest, in the different shades of red, is broken up into seven marketing boards...

06:07...and in those marketing boards are 30- to 40,000 woodlot owners.

06:12Now, I mean, I could speak for a day about how this complex ownership makes things challenging...

06:18...the competition and stumpage, all that stuff. It's difficult.

06:23But our ownership is complex, and our forest is complex.

06:26We occur, in the Acadian forest region - it's a transition zone between the boreal softwood forest to the north of us...

06:33...and the deciduous forest to the west and south of us - and it's really a mixing ground.

06:38You get about 20 different tree species and not...

06:42You know, it's 20 tree species that all have a certain amount, or a pretty good portion of, the inventory.

06:50So there's a lot of minor species; there's 20 major commercial tree species we have to deal with.

06:55And those occur in conditions from pure softwood all the way to pure hardwood and every mix in between.

07:00So it's a complex forest.

07:04Our industrial infrastructure. Six pulp and paper mills, about 14 SPF dimension mills...

07:10...and those mills consume about 85 percent of the fiber in the province.

07:16Then we've got about 35 smaller sawmills - cedar, hardwood, white pine; some panel mills; and a few pellet mills have started out.

07:25That's a pretty quick overview of the forest sector in the province, but the message there that I'm trying to portray...

07:31...is it's complex, or at least we think it is.

07:36Now, back to the challenge. Remember I mentioned in the second slide the sort of chinks in the armor...

07:41...the things that we think are shortcomings in our management of the forest or in our understanding.

07:47Supply and then the consumption requirements, the demand, and then location.

07:51And not supply in terms of total supply, but supply of different products. I just want to further that a little bit.

07:58So we don't have any sort of within-forest natural competitive cost advantage.

08:04Our forest is relatively slow growing; we grow at about 1.5 to 2 cubic meters per hectare per year.

08:11And we generate a lot of products.

08:12When we harvest a hectare, it generates up to 20 different tree species and many different products.

08:17Now, you might say, Well, that's an advantage.

08:19That's only an advantage if your collective mill demand matches what the forest serves up.

08:26If not, you have incremental costs dealing with those other products.

08:31So there's some optimal demand portfolio that has to be found there to have an advantage there.

08:38And that's really difficult with the analytical capacity that we had a year ago, that it's really hard to find that perfect...

08:45...you know, where the right mills at the right scale in the right places consuming the right products.

08:52And then location. Forestry's, at least in New Brunswick, is different than any other manufacturing sector.

08:58The more input you buy, the higher the price of the input, 'cause we have to go further for that input.

09:04So you're going into someone else's wood basket, your price goes up and so does theirs.

09:08So location is very important when we're considering this, because anecdotally, there's supply and deficits in different regions...

09:16...and that stuff's very difficult to understand if you don't have the tools to understand it.

09:21So, as I said a moment ago, the recent sort of downturn, especially in the forest economy, sort of highlighted our shortcomings.

09:29It highlighted the requirement to understand the economics of how the wood flows in the province...

09:35...where there's surpluses, where there's deficits. And there wasn't one...

09:39...you know, you couldn't say that one person sort of understands all that or one unit understands all that.

09:44There was no one big picture view to understand those linkages and opportunities.

09:50So about a year ago, there was some sort of high-level meetings, and it was decided that we needed to take advantage...

09:58...of the history we have in optimization and modeling to try to build a model or some models...

10:04....a series of models, to understand this sort of economic ecosystem.

10:10So the solution was, after those high-level meetings with industry and government...

10:14...that there would be a committee formed, multi stakeholders...

10:19...government, industry, the private woodlot owner organizations sat down and said, Well, what are we going to do about this?

10:26It didn't take long - we've worked for a long time with Remsoft - to suggest that they should be hired...

10:33...and act as a model builder, but also, we're dealing with a lot of proprietary information...

10:39...information that people weren't really comfortable with sharing.

10:41I mean, it's hard enough to get those three groups in the same room, you know, having a similar objective...

10:47...let alone sharing a whole bunch of information.

10:49So Remsoft was hired both as a model builder and as a third, trusted third party, to hold that information...

10:55...do some averaging so that the proprietary nature went away from the information, and built it all into a model.

11:02Now, I'm going to talk in a lot more detail. This is a little bit of a mouthful...

11:06...but they've created a high-level, strategic, provincial-level model of wood supply and demand.

11:14So all sources, all sinks, both internal within the province and close export locations and import locations...

11:20...and really, the big important part of that model is the delivered wood cost, the haul distances.

11:27So that's a big part of it. And like I said, I'm going to get into more detail in a moment.

11:34Okay, so four pillars. I could go into a lot of detail here; I'm not going to.

11:39I'm going to hit on the sort of four main pillars of the model, I think, just to give you an idea of its construction.

11:45And I'm going to get into a little bit of detail, some of the neat things I think we used or did with the tools to set up those four pillars.

11:53So just really quickly, for supply, we had 50 roundwood products, 25 residue products.

11:59Just to give you an idea of the complexity.

12:02We had 800 sources of supply, so we broke the province up 800 ways and said that's a source...

12:09...and defined the volume that was available in that source.

12:12We had a hundred thousand kilometers of paved and forest roads that we networked...

12:18...and then a hundred destinations, so 80 domestic mills and 20 export mills.

12:22And the tools we used to set this all up and build this model...

12:26...we used Remsoft Woodstock and Spatial Optimizer to do the supply, Esri ArcMap to do the sources and the network...

12:36...and then Allocation Optimizer was really the thing that housed all this and was the model that was delivered to us.

12:42So I'm going to go through each of these in turn just to give you a little bit of detail and idea about how they were built.

12:49So the supply wasn't new to us. I said that we've done some work on products.

12:54We did some fieldwork, got some guys out in the woods cutting trees down, slashing them up...

12:58...trying to figure out our product profile, something we haven't done a lot of in the past.

13:03I'm not going to talk about that too much.

13:05But anyway, the harvest blocks that we defined, they're in red here; this is just one part of the province.

13:11You can see the ownerships in gray, blue, and yellow.

13:14We used Remsoft Woodstock to do that - that's something that's not new to us - and Spatial Optimizer on Crown land.

13:23See, on Crown land, we've got management control.

13:25A harvest schedule can be implemented the way that it was given to us out of Spatial Optimizer.

13:31So on Crown land, we used Spatial Optimizer; we optimized the location of those harvest blocks.

13:36On private land, we don't have that control; there's no one person controlling the implementation of a harvest schedule.

13:41So we use a Monte Carlo simulation approach, which is available right in Remsoft's Woodstock, which is kind of nice.

13:48And everyone agreed to that being the best approach.

13:51So that's our supply.

13:54So sources. Now, we needed to define sources, because we couldn't have from each harvest block the distance to every mill.

14:01It just would've blown the thing up too much. So we grouped harvest blocks.

14:06And we had some interesting discussions about, well, we've got seven million hectares. How do we group it up meaningfully?

14:12And we talked about road sheds, and we actually did a little mathematics here to develop these road sheds.

14:18And I'm not sure if that's a new thing; maybe that's something that's talked about down here...

14:22...but it was new to us, and we thought what we did here was kind of neat.

14:25So I'm going to talk about it for a moment.

14:27But essentially, how does a road system capture area, just like a watershed would capture water?

14:34I used the Spatial Analyst hydrology tools when I was doing my master's a little bit...

14:39...so I was familiar with them and thought, well, we could probably use those to define road sheds.

14:44So the harvest blocks here are in red and the road in black.

14:49So essentially what we did was we turned the roads to a raster and also just a polygon of the province into a raster...

14:58...and we defined pour points. So pour points are just where one river meets another.

15:03Well, we defined them as where one road or where forest roads meet paved roads.

15:08So that's where the wood leaves the woods and gets onto paved roads, and so that's what we defined as pour points.

15:15And this is all just right in Network - or sorry; Spatial Analyst.

15:20Then we used cost distance from those pour points to basically give a height to the road and to the land between it.

15:28It's kind of like a DEM, right? It's distance from that pour point.

15:31So the further you are away from the pour point, the higher you are.

15:35And then we just used flow accumulation and direction and then the watershed tools to define where those road sheds were.

15:42So the area here, it would make more sense to go to that pour point than to that one.

15:52We thought it was kind of neat, and it worked out quite well.

15:58Okay, so we've got supply, and we've got it defined by sources, road sheds.

16:03Now we need a distance from each one of those road sheds to each mill.

16:06So the committee thought that distance and time were important from, as I said, each road shed to each mill.

16:12We wanted to leave all opportunities open, and we used OD cost matrix and Network Analyst.

16:19I think that - is that origin-destination cost matrix? Is that what OD stands for? I believe so. Yeah, yeah.

16:25Okay, so what we did was we defined a centroid for each road shed and then defined the distance and time to each of the mills.

16:35It builds a nice matrix, just like it says it does.

16:39There was some talk around the committee that centroid wasn't the best way to do that, so just...

16:43...this is sort of academic, but I said, Okay, well, let's try randomly placing points in the road sheds, one per every...

16:49...I think we did one per 2,000 hectares...

16:51...and then did the distance from every one of those points to every mill and then averaged it.

16:56Same answer, exact same answer.

16:57So if you're ever faced with that question, just use a centroid.

17:02So there we have a supply, a supply with each source, and then a distance from each source to each mill.

17:11So demand, that's the next thing. So what's the pull for that wood from each of those road sheds?

17:16So what we did is we surveyed each fiber-consuming facility in the province, about 80 of them...

17:23...and we asked them the specific product requirements that run through their mill...

17:28...so specific top diameter for sawmills and what are the lengths of products?

17:32Are you a stud mill, are you a sawmill? Tried to keep it as specific as possible.

17:38We asked their maximum capacity under one shift or two shifts.

17:41What are your residue factors?

17:43So for every cubic meter you consume, how many chips, how many shavings, how much sawdust, how much bark?

17:51So, you know, the full spectrum of things that would come out the end of that mill.

17:54How many people do you employ? So pretty detailed.

17:57And people were willing to share once they understood the benefit of sharing.

18:01And then all that information was loaded, as I said, into Allocation Optimizer.

18:06Okay, so that gave us this - we were calling it the New Brunswick wood flow model.

18:13So just as an example - well, before I get into the example.

18:16So the wood flow model, it's an optimization model, so essentially what it does is it finds the best way to go pull wood...

18:23...from one of those sources to a mill based on the products and the amount of those products that a mill needs.

18:30So as an example, so it really drives on haul cost. I said that earlier, but, I mean, once you...

18:37The input to this is a harvest blocked schedule, so harvest cost doesn't change too much...

18:47...stumpage doesn't change very much, loading cost doesn't change, so it's really the haul cost that's changing.

18:54So what we put in there is gate price, or what the mill is willing to pay at the gate, and then all the costs to get it to gate...

19:02...the stumpage cost, the harvest cost, the loading cost, and the haul cost.

19:06So for this example, spruce fir sawlogs from Crown license 1, road shed number 5; it was a clear-cut stand that was managed...

19:14...so it was a plantation or precommercial thinning. This is the profile of those, of the gate price and those costs.

19:20And what we're calculating is this residual value; that's the 925.

19:23You minus all the costs from the gate price, and you get a residual value.

19:27So because it's an optimization model, we can maximize that value.

19:32Okay, we can say, you know, maximize the - essentially, I mean, you could call it net revenue delivered to each mill.

19:39And really, as I said, it's that haul cost that's the driver there.

19:42I mean, the harvest cost is more, but the haul cost is really the thing that's changing.

19:48So, you know, if you have to go twice as far for your wood...

19:50...you end up with a negative 75 cent per cubic meter residual value there.

19:55Now, it's highly, highly powerful to constrain the model to say no negative deliveries; it has to be a positive delivery...

20:04...and then we can really understand where there's surpluses in the woods...

20:07...or where wood's not moving because it's uneconomical to do so. Highly powerful.

20:14Now, tables aren't the best way to show this stuff, and I didn't even fill this one out because we haven't nailed everything down yet.

20:20This is still a little bit - it's been delivered to us, but all the inputs are a little bit a work in progress.

20:25But now we can simplify some of that complexity I was talking about earlier.

20:29Species along the - as rows, ownership as - sorry. Species as columns, ownership as rows here...

20:37...this is something that we've never been able to fill out in this detail for the province.

20:41So the supply by species by ownership, the volume processed under any given scenario with this model.

20:48So, you know, you minus the first table from this table, and you start to understand supply and demand.

20:54Mapping's way better. I mean, this is an Esri conference; you got to show some maps, so...

20:59And it's much more powerful when you map this stuff, so we can look at any one mill.

21:03This is a sawmill in the Northeast. Where's the draw?

21:06So the darker blue is the more volume that they processed, and, you know, the mill actually sits right there.

21:15So they're drawing wood closest to their mill; makes sense. That's not particularly a surprise.

21:21This map is where we're starting to really add value.

21:24So this is undelivered hardwood species, so this is wood that's in a harvested plan but it's not getting delivered with this model.

21:33So as you can see, there's, you know, as you get darker blue, that's more volume that didn't get delivered.

21:38So you can see the center of the province is Crown land, there's not a lot of roads in there; a lot of that didn't get delivered.

21:44Up until today, this is just anecdotal information.

21:46People say, you know, hardwood's really not moving from the center of the province. You know, you can't confirm it?

21:51This is so valuable to sit down in front of people and start talking about it.

21:55I mean, we can develop relationships now with our economic development agency, you know...

22:02...true investment opportunities and things, so highly valuable.

22:07We're looking at a new policy for clear-cutting, trying to reduce the amount of clear-cutting in the province...

22:11...and we can start to look at what's the impact of that policy in wood movement in the province.

22:16So I'll just go through this really quickly.

22:17But you can see, you know, by relaxing that policy, the undelivered wood, those road sheds get lighter in color.

22:25They go from darker blue to lighter blue.

22:27So you can see, you know, you're not changing haul costs so much, but you can go further because the harvest cost is lower.

22:33So interesting policy decisions can be made from this model.

22:39So for moving forward. This is our first big step in trying to model this economic ecosystem in the province...

22:48...and we're really excited about it.

22:49It's just like the early '80s. You know, we've got a process in place now...

22:53...and we're just going to keep moving to improve the information and to build that analytical capacity.

22:59But we feel as a province we're really breaking ground here.

23:02And down the road, we want to add some of the downstream economics like jobs and shipment values, GDP, taxes.

23:10That type of thing would be highly valuable for our province.

23:13So the collaboration has been key so far. To get those people in the room together to decide on a project moving forward...

23:20...and to collaborate with Remsoft, it was a real pleasure, and it's been a success so far.

23:25And that collaboration is required to move forward, so...

23:30If I have time, I'd be happy to answer any questions - which I don't think I do. Okay.

23:34Well, thanks, very much. Thank you.