Video

LLamasoft & SCDigest Achieving Supply Chain ‘ESP’ with Enterprise Simulation Planning

Video Transcription

Dan:
In an age of growing volatility and incredibly dynamic supply chains, wouldn’t it be nice to have a little bit of ESP to know what was coming at you and how your supply chain was going to perform under different conditions? Well, I don’t know if you can get extra sensory perception, but you certainly might be able to benefit from enterprise simulation planning, another form of ESP, and that’s what we’re going to talk about today here on this video cast. It’s all about what simulation and the supply chain can do. Simulation in this context is a cousin of the network planning and optimization design tools that we’re well familiar with, but we’re going to learn today about a companion technology and approach around network simulation. I know you’re going to enjoy this broadcast. I’ve learned something already just from the prepping for it and we’re all going to benefit from the insights today on that.

One of the reasons I know we’re going to do that is because we have Toby Brzoznowski from LLamasoft Business with us today. He’s a world class expert on these technologies. He’s been with us before, one of our favorites guests. Toby, thanks so much for joining me today.

Toby:
Thanks for having me.

Dan:
Okay, I’m really looking forward to this discussion. Like I said, just you and I interacting before we started here I got some insight already. I’m really interested in what you have to say. Before we do that, just a few quick housekeeping notes. I think we’re going to get this broadcast done in about 30 minutes or so. We’ve been pretty good at hitting those targets of late. Plan on a relatively short but concise broadcast here today. Somewhere on your screen, you may have to scroll up, you might have to scroll down, you’re going to see a chat type window. That’s how you can ask a question. At the conclusion of the formal part of the broadcast, we’ll go to that live Q & A and we’ll take as many of those questions as we can, spend 10, 12 minutes or so answering as many of those questions as I said we can do to.

Within 24 to 48 hours, you’re going to get an email from us. If you’re registered today on an On Demand version of the broadcast, we’ll have that Q & A of course with it. We’ll turn the Q & A into a podcast because a surprising number of people seem to enjoy downloading those podcasts. We’ll have a copy of the slide materials today. There’s a white paper on this topic. I’m sure we’ll make a link available for that as well. Maybe throw some other goodies as well.

I want to think LLamasoft providers, again, of network designs simulation solutions for making this broadcast possible today. Can’t do these kind of things without them and want to thank them for their support.

Toby, I’m really excited about talking about this today. Simulation has been out there for a while. You guys have certainly been known for it, but I still think the overall market doesn’t understand it well. You said that to me again when we prepared for this that a growing number of companies are doing it, but I think it’s still a somewhat unknown. It may be a little bit mysterious technology. We’re going to learn a lot today from your insights and what you’ve done here. ESP, I like the concept. What really is the opportunity here?

Toby:
Well, they’re rhetorical questions, but what if you could actually predict what your supply chain would do in the future? What if you would know exactly what was going to happen if you made a strategy change before you actually had to do it in the real world? That’s the question. What if you were to know ahead of time and be able to to predict when you would stock out or when you would have late deliveries or hit capacity constraints? That’s really the opportunity that simulation gives you that really the other technologies out there, the optimization technologies don’t necessarily provide.

Dan:
That’s very powerful, obviously. What can really be achieved? Some people probably little doubting Thomas’s out there. What can you really do with simulation technology?

Toby:
Yeah. I guess one of the things is you’ve got to think about what is simulation and the idea is with simulation you’re really just running a clock. You’re actually saying let’s start the clock and let’s run that thing and let’s watch every transaction as it happens. When an order is placed, who sources that order? When that order is sourced, how much inventory do I have there and at what point do I need to reorder inventory? What transportation mode do I use? If you could actually have all those policies and all that created in a model in a digital environment and run that out, you would see everything that happens along the way. Where optimization is is more of an aggregation and looking at the flows and trying as best prescribe for you what is the best from all the alternatives. A simulation is really doing something different in telling you exactly what’s going to happen along the way.

For us, we spend a lot of time and a lot of effort being able to make something that can scale to the scale of an enterprise supply chain for a large global 1000 company. Over time what we’ve learned is there’s a lot of different ways where simulation can be used. What we try to do here with this approach and this presentation, this framework is to help people get their head around it, get their arms around what is the opportunity with supply chain simulation.

Dan:
Yeah. Do you model the network in the same way for simulation as you do for an optimization?

Toby:
Yeah. What you do is you start with the same building blocks. You have products. You have sites. You have demand. You have sourcing policies. You have inventory policies. You have production schedules and strategies. You have transportation policies. All that stuff, the data that goes into any kind of a supply chain model, whether it be around inventory optimization or network planning or product flow path analysis or enterprise simulation, there’s all the same data. That’s the good part.

The difference is you’re going to get into usually one more level of granularity with simulation than you do at an optimization level. You’re looking at every individual transaction and you’re looking at every individual SKU. You see exactly what’s going to happen along the way.

Dan:
Yeah. Just to be clear on this for everybody. A network optimization, you do not play through all of these transactions, all of these details. You’ve got to get an answer. You can run scenario analysis or whatever. That really is the core difference?

Toby:
Yeah. We often say the difference is optimization is a prescriptive technology. It’ll take all of these sometimes hundreds of thousands or even millions of possible structures and it’ll identify the lowest cost, if that’s your objective, or your highest profit solution, but in doing that what happens is you’re optimizing against a time horizon. Sometimes it’s over the course of a year or a 5-year plan broken into yearly time buckets. Maybe it’s a 1-year plan broken into quarterly or monthly time buckets. The challenge is that within that time bucket that you’ve created and let’s say it’s even a month, you don’t see every individual transaction that’s happening. What is does is it aggregates the flow throughout that. You never know that on a Tuesday of every month that’s where I have the most constraints because I have to make all these deliveries that have been made and placed over the course of the weekend or whatever that might be.

Where simulation is a little bit different. As opposed to be prescriptive, we call that predictive or descriptive technology. It takes one specific scenario, it doesn’t evaluate a bunch of alternatives, it just looks at one specific scenario and tells you in a lot more detail exactly how that scenario is going to play out, every individual transaction along the way. You can see really clearly all the time series information on how much inventory you always have on hand or where are you stocked out, how many orders were late, how many were on time, how much was even on the trucks. That’s the kind of detail that you get in the simulation.

Dan:
Another thing in prepping as I sit here for this discussion, one of the things you brought up was the advances in the computing power in memory, all that stuff has really been a key factor in enabling simulation to go to the next level in supply chain.

Toby:
Yeah, exactly. What we’ve seen, this technology simulation has been around for decades. It’s not something new. The real challenge was in the ’80s and ’90s and even the first part of this century is you were limited in a number of different areas. One was memory within computing environment. You had a 32-bit operation system. You probably had 2 gig of memory to play with. If you’re a company like a global or a national grocer that has millions of demand transactions and tens of thousands of SKUs, you create a simulation of that environment. You hit that memory constraint, you’re pegged out. You can’t go any further. When you move from 32 to 64-bit operating systems, immediately now the memory is unconstrained. You can go as much RAM as you want.

There’s other factors like memory databases that make all that data hot and ready for you. You have cloud based computing which is another thing where if I run dozens of different scenarios or maybe my machine itself doesn’t have the bandwidth to run one of these models, but if I have processing power on demand that I can get in the cloud and I can run tens or dozens of scenarios or run really big models on processors when I need them, those kinds of things have just come about in the last couple years. It’s really changed the whole landscape of what’s now possible in simulation.

Dan:
Yeah. It’s really just opened up a huge opportunity to look at the insight and the information in just ways that were so constrained before. Now you can do almost anything really which is amazing. You and I have talked about it in network design too, building this uber model that maybe just wasn’t possible in the past and now can be done.

You guys have had this technology for a long time. Maybe the latest breakthroughs are now on the technology side, as you said, some other factors, growing interests in your customer base. You super charged what you’re out there doing with the simulation and you’ve come up with this concept ESP. I like that term for reasons I said. What’s that all about?

Toby:
Yeah. Really simulation is in many cases extra sensory perception for the supply chain planners. Giving you that ability to predict what will happen in the future, but the challenge that we’ve seen is how do you wrap your head around what are the different ways and use cases for this. In doing that is where we identified a couple of the variables that are most changed when you’re trying to evaluating these different options and one of those is you could simulate and run models about your existing network with your existing structure and your existing policies or you could be looking into the future of different policies, different strategies, different ideas. At the same time, you could be looking at the demand that’s occurring today, the last 12 months’ worth of demand that I had to fulfill, or I could be looking into the future of demand that hasn’t even happened yet. I’m predicting new products or new geographies or the combining of two companies as I go through an acquisition or a merger.

The network structure piece and the policies are one of those dynamics and the second one was the state of the demand. Is it today’s demand or is it future demand. When we started to look at that, these 4 quadrants emergent and that was the 4 quadrants of ESP that put the framework together we’re talking about today.

Dan:
Yeah. That really gives you the 4, at a high level, the 4 use cases. These are the 4 different ways and you could use all of them, a given company could use all 4, but these are the 4 different ways companies can benefit from simulation type technology. Why don’t we go through the 4 that you’ve got here and the first one is baselining which is the current network and current demand.

Toby:
Again, you think about these 4 different states, current network structure, current policies and policy could be something like my production process or it could be who sources which store. I’m a retailer and I’ve got 10 DCs across the nation and I’ve got 5,000 stores and each of those DCs has to service a certain amount of products to each store. Which one is connected? It might just a flow path of how I flow products in from my suppliers into the network. You try all these different policies or network changes. It could be today’s or it could be the future and at the same time, the demand, current or future. That first quadrant is I’m looking at my existing supply chain with the existing policies and I’m looking at the existing demand. What does that do for me?

We call that baselining and whether you’re doing network optimization or you’re doing simulation, you’re doing any kind of modeling and analysis, you need to start with a baseline of model that represents your existing structure. Creating a simulation that basically takes, let’s say, your last 12 months’ worth of history of all your shipments and orders that you had to fulfill given the most current network structure, you create that simulation, now you have a baseline. You basically run the network. You’ve run all these simulations and what you want to do is validate this model that I created actually represents what happened in the real world.

I’ve talked about this in the past in projects. Once you get that validated baseline model, that’s the part where you party. You throw a party, everybody goes out, has a few drinks, you show up late the next morning because you’ve accomplished the biggest milestone of the analysis because that becomes the launching pad for all the future things that you do. What’s cool is there’s also some benefit you can get just by creating the baseline. A lot of times you can just now visualize what is going to happen.

Visualize is one of the things. If I can see exactly how all the products are flowing through the network and I can see where I’ve been hitting capacity constraints or where my orders have been late, just by being able to visualize that, sometimes it’s hidden in all the data that you have and all the data that you have to deal with. Creating that visual model, that representation of it, gives people immediate insights into what’s happening that you can fix quickly. It also gives you another mechanism to communicate what’s actually happening in your supply chain to the rest of the organization that maybe doesn’t understand all the dynamics of the supply chain. That baseline is a key and, like I said, it’s the launching point for all of the other analysis that you do.

Dan:
Yeah, very cool. Like I said, there’s so much happening with large, complex supply chains today, I’m sure it’s so easy to miss things that are happening out there and you’re actually taking real data, running it through real supply chain analysis and now all of a sudden these insights pop up and, as you say, great starting point to move around the rest of the quadrant.

Why don’t we go to the next quadrant which is quantification which is, I think I got this right, current demand, new network design. Tell us about that one.

Toby:
Yep. Let’s stick with the last 12 months’ worth of orders and shipments that we had to make and everything, but now let’s think about what if we changed our supply chain. What if we hadn’t been structured this way? What if we had tried a different replenishment strategy? Maybe we tried twice weekly deliveries instead of once a week or what if we had 5 DCs instead of 4 in this region or what if we flowed products in through the port in the Gulf, as opposed to the east coast and the west coast? What if made these changes to the network, future theoretical network? That’s this quantification piece because what it does for you is basically allows you to validate decisions ahead of time.

I have an idea that if we were to make this change, it’s going to make an improvement. I have a gut feel and you have executives all over the place that they just, if they’re the loudest speaker in the room a lot of times that’s what happens. You want to be able to have a little bit more data driven approach to decision making and this is one of those ways to do that. Before it happens, I’ve got this baseline that already shows what we’ve done in the past. Let’s just make a change. Let’s change the policy. Let’s change the structure and let’s simulate that same demand and see what would have happened.

You might find that you can reduce your cost immensely with that new structure, but there’s going to be a serious impact on service if you do that. There’s stuff like that that you can validate before it happens and that’s also a cost avoidance situation. Before I implement this new policy or this new strategy, let’s test it out in the digital environment.

Dan:
Let’s go back to what you said a second ago to make sure I’ve got this. You could get some of that from network optimization, but as you said it’s going to aggregate the data over some block of time. What simulation’s going to give you is where there might be problems or more opportunities, I’ll stick with the problems, on a discreet basis. Even the month or the quarter looks good, I’m running into certain problems every Monday or whatever.

Toby:
Yeah, you said it. Let’s just use a simple example of I went from 4 DCs to 5 DCs. Network optimization tells me that’s a lower costing way to do it. Here’s which DCs should be assigned to which customers. Here’s how much in aggregate is going to flow. Maybe it’s over the course of last year if we would have done this versus this. You simulate that and what you see is I’ve created a bunch of new lanes. I’ve got different volumes flowing through these different facilities. Where before, I was able to run full truckloads. Suddenly some of these trucks are half full. I’m not getting the same cost that I think or it is that on Mondays and Tuesdays on these first week of the month, I’m having these huge volumes and then on the rest of the month it’s trailing off. You can’t see that if you’re just running an optimization. You don’t see that level of granularity and all the details. Where here you might see I’m hitting huge capacity constraints that are causing me to actually miss shipments and causing me to have to expedite orders and the actual cost is very different than what the optimization gave you.

Dan:
Very good. Well, let’s keep moving on the quadrants here. Next, I believe the quadrant is forecast which is current network, future or change demand. What’s that one about?

Toby:
Yeah. The other thing that’s been really cool is now if you have an existing structure and a model of your supply chain, there’s always new forecasts coming in. There’s always new products being introduced. There’s new potential geographies that we’re starting to serve that changes the dynamics of the demand. One of the processes that’s starting to be used more and more is using simulation in your SNLP process. I’m getting this new demand forecast over in my capacity planning process. Let’s simulate that and see what’s going to happen. What’s going to happen over the next month or 2 months? Is there a situation where I’m going to run into an issue where I have to maybe smooth out demand, where I have to preposition some inventory or I have to prebuild product? You don’t necessarily know that unless you can simulate and test those things out.

Going ahead, you’ll hear people calling things like flow casting or future casting in the network. Those are some of the things that you can do if you’re simulating ahead of time and running these simulation in parallel with your operational plan. You’re not changing the structure. I’ve got the structure. I’ve locked that in, but what I have is different demand.

The other part is I could run sensitivity around that. I could be testing out and saying, “This is a great structure and this is what my demand has been historically. What is my demand goes up by 5 or 10 or 20 or 30%?” Now how is my supply chain going to perform? What if it goes down in the other direction? How is going to perform? You can use this kind of forecasting structure here to test out the sensitivity of your supply chain with those variables.

Dan:
For example, because I’m trying to get my head around this, maybe in one of those scenario planning’s or analysis this shows me that I’m going to be out of stock on or all told I’m going to be hitting my service level targets, but once a month on these products I’m out of stock. Then my decision is can I live with that and I’m going to be okay with that or do I need to make some changes so I eliminate those things? You’re seeing it an on individual basis, not lost in the average.

Toby:
Right, exactly. That is the difference and what’s cool is you can be running these things even if you’re looking ahead 3 weeks or a couple of months ahead of time and constantly running simulations as you’re doing your business and it will pinpoint those areas where I’m not going to be surprised if I have capacity constraint at this DC 3 weeks from now. I’m going to know a little bit ahead of time what’s going to happen.

Dan:
Very good. Okay, let’s go to the last quadrant which is innovation and I’m looking at new network, new demand. That’s going to be an interesting sandbox to play around in.

Toby:
Yeah. This is the design world that we’ve always lived in at LLamasoft which is let’s just see if we can make changes to the network and we can imagine something new, what’s that going to look at and how is that going to make the big step change that we need. What we’ve seen is there’s a couple of different areas. There is obviously that strategic planning which is I’m considering something like an acquisition or a merger or I want to go into a new market. I want to introduce these new products. There’s all of these different strategies that you’re considering along the way. Well, what’s going to happen? In a lot of cases you can’t go back to historical demand because you didn’t operate that way. You didn’t have those products. You didn’t work in those geographies. You’re not only trying to predict and look at a new structure, but you’re also doing it with theoretical demand. You want to run a bunch of different simulations under different assumptions to see how is your supply chain going to perform in those environments?

It’s a compliment to an optimization approach which says here’s the new potential structure that an optimization would prescribe. Now let’s see how robust that supply chain is under these, and you can change variables. We’ve seen some of our clients where they’ll run hundreds of different scenarios. I’ll adjust my lead time by these different parameters and I’ll also adjust my production throughput and my actual yield and things like that. They’ll predict different demand characteristics and run those variables. You start to get this cloud of different results that show you where you’re at risk and where you’re safe.

The other thing that’s really cool about this and we’ve heard some of our clients do and I’ve heard I think one of your guests in the past, Jake Barr who is a former exec at P & G talk about this, is make it into a bit of game. I have people are doing the optimization of my existing structure and my existing supply chain, but let’s let a few folks go out there and really try to think outside the box. Let’s try to come up with a new concept, a new idea. Let’s break all those business rules that we’ve lived with for the last 25 years that have made the company what it is today, but maybe there’s a new business rule or a new strategy that’s going to take us to the next level. If you set a couple of people out there and make it into a game, gamification, it’s an interesting thing. Who can come with the most groundbreaking concept and let’s simulate these things and see what’s going to happen. A lot of times, those end of being the thing that take you to the next level.

Dan:
Yeah, great. For every billion you save, you get a 100 bucks or something like that.

Toby:
There you go.

Dan:
Okay, very good. Well, great description of the 4 quadrants. I love that overall approach when you’ve taken something that’s kind of hard to get your arms around and basically organize it into these 4 quadrants. Now, to take that to the next level, you’ve actually gotten a real example and we shared offline some of the actual companies, but I can’t name them here obviously. You’ve got an example of each of these 4 use cases. Why don’t you walk through those as well?

Toby:
Yeah, I’ll run through a few of these really quickly. I’ve talked about them, but just in the baselining itself. A global chemical company creates a model and runs a simulation of their existing network with their historical demand. They’ve come up with a new concept that for them, they’re really thinking about rail car utilization. How well do they utilization the rail cars that they had least last year? It’s a very expensive asset to keep on the books. The future state was really I want to test out new scenarios. I really want to see what’s going on.

The baseline itself, what was really cool about that is just by having that in place is they started to see really quickly is they were to just keep the exact same strategy but position things differently, they would get a different result because they were hitting a specific capacity. They were late in these 3 areas consistently all the time. That baseline was showing them something they didn’t necessarily know. They saw there was over-utilization in a couple of areas and under-utilization in others and then, of course, they used that as the launching pad to the next level of design, but just seeing exactly how thing flowed through the network, how those assets were utilized and where the constraints were, where the bottlenecks were in the process was very eye-opening for them. That key of using a simulation to help communicate then to the rest of the executives here’s a problem that we can deal with, here’s a way in which the current policies that we’re using aren’t working. That was a great first step.

Dan:
This visualization you talk about, my quick reaction is it wasn’t that long ago that just having a visualization of how product flowed across your network was a breakthrough for companies.

Toby:
Even a flow chart, yeah.

Dan:
Now, you’re taking it to simulating rail car use on a daily basis or whatever, just a whole new world really is to me.

Toby:
Exactly. Then take that and once you have the baseline, again, that’s the launching point. The quantification example where it’s I want to look at, again, take the same orders that I had to fill, the same shipments that I had to make, but let’s look at a different policy, a different structure. This is a global tobacco company and for them, and tobacco actually has very smooth demand. People smoke a pack, they smoke 2 packs. They buy it from the same place all the time. Demand is actually smooth, but when they started to look at what was happening in the manufacturing facilities at there different global locations, there were these huge spikes. People were running these giant batches and then what was happening was all these expedited orders were happening out to the suppliers. The suppliers were frustrated, but the costs were higher than they need to be. There were these situations where they suddenly had over capacity of some things and then they had these quick runs that they had to make because they were making their delivery schedules.

The guys at the corporate headquarters had this idea and said, “Look, we know what the demand looks like. Let’s simulate a different policy, a different strategy for how we’re going to do our production planning. We’re going to segment our products into 4 different categories. We’re going to run some on a weekly fixed schedule, some on a biweekly, some on a monthly fixed schedule. We’re going to take these low volume products and we’re going to have a fixed amount and we’re just going to run that when we need it. Let’s look at that last 12 months and let’s pick one of our facilities in Malaysia. That’s a big problem area for us. Let’s go back over the last 12 months and what if we had actually run our supply chain with this production schedule and not changed it at all? Just run it as we’ve written here and run it. What would happen?” It was phenomenal what they saw.

Basically the expedited orders, they went away. The amount of on-hand inventory was reduced by almost 50% and the in-transit was just so smooth that they could predict exactly what was happening. The cost, it was tens of millions of dollars a year that they were saving by running this. They used that to go back to the plant, which was making money for them, it wasn’t like it wasn’t profitable which is one of the reasons why they didn’t want to change anything. They said, “Look, if we make these changes, here’s what we’re going to see happen and by the way, your production planners are no longer going to be making these on the fly changes. They have to run with it.” They ran the pilot for 3 months. They tested it against what the simulation was and they found the results were right in line with what they simulated. They were able to roll that out to not only that region, but other regions and it became a new plan.

Dan:
That certainly sounds like some darn good ESP to me in your supply chain. Very good proof point there. Okay, on the forecast for the pharma companies using the tool for some SNLP supply.

Toby:
Yeah. They have advanced planning systems. They have SAP. They’re using APO for their production optimization, but what they wanted to see is as these new plans came in, there were still situations where they were missing orders, where they were having issues. They essentially emulated the process in the model that they were using in that APO module over here in a simulation. Then what they would do is on a monthly basis, as the new demand forecasts were coming in, they would have an optimal solution that came out of APO, but then they would simulate that optimal and they would see what’s really going to happen. They would find a lot of situations where there would be these certain days or weeks or certain products that weren’t going to perform as well as what they had predicted because they were looking at it at the next level of granularity.

Then they were also using that to test out what if we make a change here? What if we make a change here or what if the demand isn’t exactly what’s forest because forecast has error? It has a lot of error sometimes. What if it’s off by 10% or 20% in either direction? How, let’s say, at risk are we to running into an issue? They would use that simulation to validate their planning process. In many cases, they’d come up with a slightly different plan than what came out of their APO module.

Dan:
Yeah, yeah. The SNLP process generally has been increasingly technology enabled over the last 5, 10 years, but this now adds a whole other dimension to it in terms of insight you can actually simulate what the effects of various decisions, whatever. Very, very cool. I love the way you’ve got these examples here because you’ve got 4 very different industries here on each use case. The last one is a global retailer.

Toby:
Right. A company with thousands of stores, dozens of distribution centers and for them, they really wanted to test the effects of different sourcing strategies that they were thinking about making. Also, just different routing decisions. If we were to try a different transportation route, you have a lot of these multi stop routes that you make to your stores, different delivery window constraints that these stores basically put on the DCs and put on the transportation network. Different replenishment frequencies, all of those things. How is this going to impact our service? Then these guys also dealt with eCommerce and they had a home delivery network as well. They weren’t just delivery to stores, but they also had routes that were going to individual offices and individual consumers.

Being able to test out what’s going to happen if we implement this new strategy at a detailed level? Is that going to impact all of my deliveries? How’s that going to impact the cost? How is that going to impact the amount of assets potentially that I might need? Do I need to add more drivers or add more trucks in these different locations? It’s a challenging problem and it has be done at an SKU level and at a transaction level of detail. You’re starting to talk about millions, in some cases, of monthly orders.

For them, it was the mother of all simulations. They hadn’t considered it before, but once you create that model of your network, now the insight that you get into these changes is so revealing as to what’s actually going to happen that they didn’t have to test it out in these little pilot projects and say, “Let’s run it for 2 months before we know what’s going to happen.” They could run it in a simulation and predict what’s going to happen.

Dan:
That’s pretty powerful there, again ESP.

Toby:
Exactly.

Dan:
Those are 4 great examples and everything. Nicely illustrated the 4 pieces, the quadrants that you have there. It seems to me as we’re talking here and we talked about it, obviously, earlier is there’s 3 technology pieces. There’s the simulation tool itself, capabilities you bring. We talked about the computing and database and all the other things were happening to enable much bigger data sets to be simulated or whatever. There’s also the data availability and the quality and this has been a bane of planning and network design and I’m sure simulation as well over the years, but that is getting so much better now with the ERP systems or whatever. That’s the third technology piece, isn’t it, that’s allowing us to come together and really drive ROI?

Toby:
Yeah, no doubt. We probably think about supply chain too much and come up with acronyms like ESP and enterprise simulation planning and things like that, but that’s the world that we live in. One day we were sitting around with these ESP and we were combining that with all of the data that’s in ERP and we came up with this formula internally that we use inside of LLamasoft that ESP plus ERP equals ROI. It stuck and we’ve used that for years internally just to drive home the point. As we were putting this framework together, we decided if we’re going to communicate this to somebody, here’s what they need to understand is that you can create these simulations, but it is driven by the ERP data.

If you can create simulations that are automated in a fashion that you don’t have to spend a month or 2 months or 6 months building a model of your supply chain or by the time you’ve built the model the problem that you were trying to solve has long since vanished or changed, if you can automate that process from your ERP data, you’re going to get a much bigger ROI. That is a reality and that what we’re seeing our customers starting to do more and more is automate the process of taking that data that’s available now out there, automatically creating these models and running them on a much more frequent basic so that they can get the true ROI.

Dan:
Yeah. There’s multiple issues. There’s data availability, can you even get the data? It wasn’t that long ago, I’m sure in many cases still today, how good is that data, how clean is that data? Then there’s just continuing advances on the ERP side and data warehousing and all that other kind of stuff that’s going on. It just makes your job and the customer’s ability to get to something of value much quicker and that’s one of the points you’re trying to make.

Toby:
Absolutely.

Dan:
Very good. Well, Toby, it’s been a fantastic discussion. Why don’t you wrap it up for us? Give us a subway view of what the opportunities really are, as you said, with simulation.

Toby:
The opportunity is you can today create a simulation of your supply chain and supply chain all the transactions that are happening. It’s going to, when you run that simulation, it’s going to give the full detail as to what’s going to happen and that is ESP for me. That’s the enterprise simulation planning. If you make that part of an integrated process, it’s not just a project that you do every now and again, it’s a business process, it can really deliver some serious results and that is what we’re seeing. The companies now that get this are using it as a competitive advantage. This enterprise simulation planning can be a pretty dramatic competitive advantage that they can use because they can predict when they’re going to have issues that maybe their other companies are also going to be faced with. If they know them ahead of time, they’re in a much better position.

Dan:
Yeah, very good. Well, Toby, it’s been a great discussion as always. Look forward to being back in the Q & A with you in just a couple of minutes here.

Toby:
No problem.

Dan:
Okay, with that. We’re going to wrap it up. That was really good. I’m sure the Q & A is going to be a fun session as well. With that, we’re going to march to that live Q & A. You’re going first see our new screen of resource, how to get a hold of Toby, LLamasoft, et cetera. We’ll have a link to that white paper I referenced. That’ll also be in the email you’ll get in a day or 2 here as well. You don’t have to madly scribble that down if you don’t want to, but now we’re going to move to that live Q & A. You’ll see on your screen a link. Just click on that link. It’ll take you to a live audio stream of the Q & A. You can continue to ask questions on any of the screens you’re on. We have that chat window on there. Just type in your question, hit submit. We’ll get up as many as we can.

Certainly thank you for joining us on this excellent videocast from Supply Chain Digest and the Supply Chain Television Channel. Look forward to seeing you on another broadcast very soon. Thanks so much everybody.

Okay, this is Dan Gilmore from Supply Chain Digest and the Supply Chain Television Channel. I want to thank you for joining us for that outstanding videocast and I promise this will be an excellent Q & A session. I thought that was just really educational and great framework for presenting what you did with the supply chain simulation. Very pleased, of course, to have Toby Brzoznowski from LLamasoft with us here today. He’s actually over at the UK at the Gartner Conference. Dialing in from there. Toby, thanks for joining us.

Toby:
Thanks for having me, Dan.

Dan:
Hey, just really quickly as I say that Gartner, I know that around the time that this videocast here that you’ve been briefing some of the analysts on that 4 quadrant framework and, frankly, I wouldn’t expect most analysts to know a whole lot about simulation. You said it’s been pretty well received. Is that correct?

Toby:
Yeah. For many folks just wrapping their arms around use cases of simulation has always been a question that I’ve gotten from them and part of putting this together was just a way to help communicate the ways in which our customers and our own consultants and people who are out there trying to solve supply chain modeling problems are really going at this. It’s been very useful.

Dan:
Yeah, well certainly I’ve not seen in all the years I’ve been following and covering the supply chain I’ve not seen anything quite like it. Congrats on a great framework. Okay, let’s get into the questions. We did get a good handful of audience questions. Before we do that, I’m going to, as I often do, start with a couple of questions from me.

Toby, it seems to me as, that discussion or whatever, I reference this in part of the broadcast, but the results that you get are often connected to either events or patterns in your supply chain execution that you just don’t see without this kind of a visibility tool. What are your thoughts on that? Am I close?

Toby:
Yeah, there’s a couple of ways to look at it. First of all, if you’re running a traditional optimization technology, what you run into is you’re averaging things and you’re averaging flow over the period of time that you’re optimizing against. You miss those individual transactions. What you find when you’re simulating is some of the spikes or some of the peaks or where variabilities tend to stack up across. Maybe sourcing variability as well as transportation variability and some of the manufacturing lead time variabilities, you see where you might run into late deliveries or late orders during certain periods of time or during certain peak ordering times during the course of a week or a month. Those are the things that jump out at you when you’re looking at the actual outputs of a simulation.

Dan:
Okay, very good. Let’s get into, oh I had a second question I wanted to ask and this is a [crosstalk 00:39:48]. As I’m listening to this, Toby, it seems to me that rather than looking at only a specific problem like the rail car example like one of the case studies or several of them, that I might want to do this on a recurring basis. Why wouldn’t I want to simulate Q1 against Q1 demand and Q2 against Q2 demand, et cetera? Why wouldn’t I set up a process where this becomes a matter of course in how I analyze and look at my supply chain going forward?

Toby:
I guess that is really what that lower right quadrant is looking at is that whole idea of forecasting and using it as part of your SNLP or your planning process. The challenge I think has been that only until recently have companies really been able to model their end to end supply chain at that transaction or SKU level of detail. There’s really a select few groups of companies that have put the process in place, adopted the technology and have really made that part of their business practice, but yes that is absolutely a great use of the technology because you can see ahead of time here’s my forecast for the next quarter. Here’s where we might run into issues whether it be hitting capacities, constraints, hitting some issues with timing or delays. Those are the things you can see pretty clearly when you simulate ahead.

Dan:
Okay, very good. Speaking about seeing clearly, a couple questions came in the end around what is the output of this look like. I don’t know if that varies by which of the 4 use cases I’m in or the specific problem or whatever, but you can clearly obviously produce these results. How do I get my arms around what does it look like?

Toby:
Yeah, there’s a heck of a lot of detail there. There’s a lot more output that comes out of a simulation than out of a traditional costing optimization or something like that because you’re getting the statistics on each individual transaction. You’re seeing every order as it hits. When did it hit? What was the decision that was made as to how to fulfill that order? What was the decision that was made as to how to ship that order? What was the decision that made in when to replenish inventory? You see a whole ream of data in terms of every individual shipment, every order, every transaction, every production process and you can then view that a lot of different ways.

The best way to really look at simulation output is in time series graphs. You can see in a time series graph exactly, for instance, what your on hand inventory is. You could see at any given point during the week or the month how is it going up, when it is going down, when did I hit that replenishment or you can see statistics on deliveries and on time deliveries and how much was in each shipment. You get a lot of statistics around those time based elements that you don’t get in other analytic techniques.

Dan:
We have time to do another question, but I just want to make a point there. I assume that means the person or the team running these simulations has to have some real skill about going through all that data and really identifying what’s relevant, what’s meaningful, what’s actionable, rather than boring the executive team with chart after chart that doesn’t tell them anything for that specific problem?

Toby:
Yeah, no doubt. What we’ve seen is really the evolution over the last 5 to 10 years of supply chain design centers of excellence. People that their job is to design and continuously redesign the supply chain. They have a whole bunch of different tools at their disposal. They have network optimization, inventory modeling tools, transportation route optimization, as well as different simulation techniques. Really depending on the question they’re trying to answer, they’re using the best tool that’s available to them and hopefully leveraging that data, leveraging some of the dashboards and the reports that they’ve created and that expertise that they’ve built over time to communicate the results and be the decision support mechanism for the company.

Dan:
Yeah. Well, that segues really well into another question. You kind of answered it there really, but I’ll ask it anyways. Is it typically the same people that are doing both the simulation and the optimization or do people tend to specialize?

Toby:
Traditionally I think that simulation technology, because it was much more buying, let’s call it, tools or tools that were built for simulation were general purpose simulation tools so you had to have a very specific expertise on how to code and create and develop a simulation model. What we’ve tried to do at LLamasoft is really just build that in as part of the same application so you’re using the same data. You’re using the same nomenclature. You’re using the same modeling techniques. You’re just basically hitting a different algorithm. You’re running an inventory optimization algorithm or running a network optimization algorithm or you’re running a simulation. It’s pulling from the same data set. What has happened I think is that those companies that are leveraging technology anyways, it’s becoming the same team and the same people.

Dan:
This is just a follow up to that for me. Maybe this is a silly question, but I’ll say it anyway. Is it always obviously that you should use optimization versus simulation or sometimes is it you’ve got to think about it a bit?

Toby:
You have to learn how to think about it honestly and that’s part of the challenge that we’ve learned over time in doing business is you have to help people learn how to think about a problem in many cases. If I could, for instance, enumerate the 3 or 4 different options that I want to compare, I want compare how is the supply chain going to perform under this set of circumstances or this set of policies versus this one versus this one? If you can enumerate those, then you should simulation each of those because you’re going to get more detail and you can compare the actual performance of each of those 3, 4, 10 different scenarios. If you have so many different options that you can’t possibly run all of those out and enumerate them out, that’s where the optimization technology comes in. Where it starts to go through thousands or tens of thousands of possible combinations to come up with a proposed solution for you. Really when you start to determine what do I know about the problem or what am I trying to compare or what am I trying to solve for, that’s where the decision comes.

Dan:
Okay, very good. You mentioned a couple times both in the broadcast and obviously here in the Q & A, Toby, every transaction, the detailed transaction. Somebody sent in a question about do I really need that each and every transaction? Can I not aggregate them into weekly buckets or something like that? Does that make any sense or do you want that level of transaction detail?

Toby:
It does make sense, but it depends again on what you’re trying to solve for. For instance, if I just want to look at the overall volume flow through facilities over the course of a 5-year or 10-year period and see when am I going to start to hit to require additional capacity or when am I going to have to make an investment in new manufacturing capabilities or whatever it might be, that might be a time when you would aggregate. In most situation, the reason you’re simulating is you want to see where the sensitivity is around behaviors on a daily, hourly basis, where you’re going to hit those peaks and valleys and really run into issues. I think in most cases, it’s easier for the modeler many times to just bring in the data at a transaction SKU level of detail, let the model run it out, as opposed to trying to figure out how do I aggregate things.

Dan:
Yeah, that makes a lot of sense. Okay, it’s getting back to the point about doing this as a matter of course and you referenced the bringing [inaudible 00:47:59]. Somebody wrote in and said, “Are you saying this tool would be used live in line during SNLP meetings or would it be used offline and the data brought in to support the discussions?”

Toby:
That’s certainly our vision is that it can become something that could be more of an online, real time decision support tool, but the reality today is I think you still have a core group of modelers and analysts who are owning these models. The way in which we’ve deployed the technology to date in the market really dictates that you come up with a few questions, people are going to run some scenarios for you, they’re going to give you the results of those scenarios. If those scenarios then beget more questions, then it goes through a second round of analysis in the cycle.

Where we’ve really started to address that over the last year and a half or so is by taking this product that we call Supply Chain Guru which is that modeling tool and incorporating SupplyChainGuru.com which is a web based platform that allows the modelers to think those models up into the cloud and then share the models through a more simple, easy to use, planner view so that I can run half a dozen scenarios on the fly, really quickly without having to know all the details of how to create a model. That really is exciting for us because we’re starting to see as you deploy that through SupplyChainGuru.com that this second echelon, this second tier, a larger population of people can leverage the models to make decisions without having to be expert modelers.

Dan:
Excellent, okay. With that, I think we’re going to wrap it up, let Toby get back to his Gartner Conference here. I’m just going to real quickly, again, I referenced in the formal part of the broadcast a new white paper on this topic from LLamasoft. You’re going to get a couple of opportunities to download that. It certainly covers some of the same ground here, but it goes into more detail in certain areas. I certainly encourage you to take advantage of that.

Toby, as always, it’s been a great discussion. I learned a lot. I’m sure our audience did too. Thanks so much for joining me today.

Toby:
Thanks, Dan.

Dan:
Okay, with that, we’re going to wrap it up. Thank you for joining us on this outstanding broadcast. Again, look for an email with the On Demand version of this Q & A, the podcast, the slides, all that stuff. You’ll be getting that in the next 24 to 48 hours. We want to thank you, again for being here. We look forward to seeing you again on another videocast very soon. Thanks so much, everybody. Have a great day.

Ready to learn more? Click to request a live demo with the LLamasoft team.

Request A Demo