KEVIN HUME
Director, Consulting Services
ESYNC
Imagine being able to predict the exact throughput of your facility, the precise times you’ll need to bring in those ten extra workers, just how much return on investment your expensive forklifts will generate two years from now, and whether a proposed change to the DC will be successful. Far from being a dream, all this — and more — can be easily achieved with a warehouse simulation program. Thanks to swift advances in simulation technology, computer modeling of facilities is no longer restricted to large industrial plants or highly automated operations. All you need are enough information to put into the model and a few weeks to program and test different scenarios. Studies have shown that simulation is a major contributor to warehouse productivity, yet direct-to-customer operations rarely turn to modeling to cost-justify facility improvements. O+F discussed the benefits of simulation with Kevin Hume, director of consulting services at ESYNC, a supply chain consulting firm based in Toledo, OH.
Few direct-commerce operations invest in simulation programs. Why?
One of the main reasons would be a misconception of the value of what simulation provides, as well as how much simulation has evolved and improved through better technology. I was involved in it in college, actually, going through industrial engineering, and we were writing programs and running simulations of material handling systems and discrete events, and with the hardware and software of the era, it took days to develop a program, and sometimes weeks to get multiple scenarios run and compile the information that you needed to make any meaningful observations.
Fast forward to today — with the computing power and technology that we have, we can put together a 3D simulation and run an entire year’s worth of data within minutes and hours on a laptop PC. Right now, I could show you on my PC a mock-up of a year’s worth of analysis and three-dimensional renderings with a fly-by wire going through the entire facility, watching color-coded boxes of different velocities moving around, and the entire development of that, from cradle to grave, would be about three weeks’ time. People just may not realize how much more effective simulation can be today, with respect to the investment and value that can be brought to bear from it.
What can a simulation program do for a direct-commerce company?
The dynamics of the business are that you’ve got high-volume product that constitutes a small number of SKUs, and a larger number of SKUs that constitute lower volume. Higher volume gets picked much quicker because you travel shorter distances to get the product, and you typically pick larger quantities. So your product is accumulating much faster, you have more of it, and you want to minimize the amount of travel that it has to take on the conveyor, because conveyor is expensive.
The other end of the spectrum is [that] you’ve got a smaller volume of product that gets placed on the conveyors, or any kind of system, much more slowly, because of the distance you have to travel. There’s a balance. So consider simulation as being a fulcrum, and you’ve got a balance of material handling equipment, conveyors and the like, on the one side, and the balance of throughput that you need on the other side. What simulation does is help define that fulcrum point that balances the appropriate amount of technology, of capacity on a conveyor, with the throughput that we’re looking at for the building.
All we’re talking about here is material handling simulation. We can also apply these same techniques to the entire supply chain, simulating [the process from] over in Asia, where the product starts, coming over to ports and fulfillment centers, then going off to regional DCs, and then dispersing from there. There’s network optimization, and there are simulation tools that simulate all the elements of the supply chain, and those also apply the same fundamentals of flow between those different elements.
If you had to pick the most important benefit of simulation, what would it be?
Validation of the original design. You can go in to the executive committee and say, “Based on what you’ve told me about how my lines per order are going to grow, and how my overall orders are going to grow, and how my SKUs are going to grow, here’s what the projected throughputs are for 15,000 orders per night, based on these projections, and this is how long I think it’s going to take to do it.”
That’s what people do simulation for — to feel good about this multimillion-dollar investment in automation that they’re making. I’ve been involved in several automated facilities that didn’t have the benefit of simulation, and what happened in most cases is that there was a period of anywhere from three weeks to three months where we spent nights and weekends looking at the results of how conveyors got backed up and why they got backed up, and waiting for the next time that particular combination of events came into play and that problem manifested itself again. What simulation gives you is the visibility to create these issues before you ever turn the system on.
Five years from now, would your projections be valid?
Absolutely. You’ll get into Year Three or Year Four, and you’ll take your actuals and compare those projections and see how close they are, and you can go in and make modifications, run that new set of data through that same simulation, and look at what that simulation tells you in relation to what’s going on. And where there are differences between the two, you can go back and look at the individual elements and begin to pinpoint what the gap is between the simulation and what’s actually going on out there on the floor.
What kinds of companies use simulation the most?
The primary users are folks that have larger facilities, most of them automated or semi-automated facilities with conveyors; users of conveyors and integrated, automated material handling equipment — horizontal and vertical carousels, tilt tray sortation units. You see it primarily in distribution in a retail environment or anywhere where you have a pretty dynamic order profile, where any element of that order can change significantly over a planning horizon. It can be because of seasonal swings. It’s a challenge to design a facility that may have to ramp up to four or five times what its average use is. A lot of times what you want to do is design that facility to run multiple shifts, and if you want to get at what those labor requirements will be and what the real top end of that facility is over multiple shifts, simulation is a good way to look at how long it’s going to take to swallow the elephant.
So simulation would be perfect for apparel or gifts retailers.
Yes — anything in retail that’s on a larger scale, where we’re looking at a significant investment in material handling equipment.
Let’s say that I’m starting a facility and I want to invest in simulation. How much can I expect to pay?
I think it needs to be placed within the context of justification. Just as when we talk about costs for material handling equipment or warehouse management software, or any other business investment, it has to be within the context of justification. So I’m going to throw out some costs, but then I want to go back to those benefits that we were talking about. We only got through that initial one, and it’s those other downstream ones that I think probably provide even more value.
What I described with that three-dimensional fly-by wire and being able to emulate an entire year’s worth of data for an expanded facility, with somewhere between two and three miles of conveyor and a conveyor control system, would come in at $40,000 to $50,000. In the simulation I just described, based on a statistical distribution, I would look at an order profile for a particular segment, say the A movers. I would pick a statistical distribution that fit the graphical representation of the SKU volume I saw. Once I’d picked that distribution, I’d say, I want cases to come out ten feet apart at intervals based on what this distribution is telling me. That’s a pretty high level. It’s good enough for this particular simulation. If we were to go to a lower level, where we would use, for example, industrial engineering time standards to define the travel time to the pick face, from the pick face to the conveyor, and from that point to the next pick, getting to another level of detail, then you’re looking at higher costs, I think. Obviously, when you look at more complexity, where we’re going into trailers and maybe we’re looking at something like tilt trays, and we have a packaging process, you would see prices go up from there.
Let’s talk about some of those other downstream benefits.
What simulation gives you are the foresight and ability to balance labor. What we’ve found is that we reach a steady-state environment, with maximum accumulation to perform sortation, at a point when the people in slow-moving areas come in 30 minutes prior to the fast-moving areas, so they gain some accumulation. Then the fast-moving people come in and begin picking, and then we bring sortation on, then we let the truckload areas accumulate for 15 minutes, say, before we bring in the truck-loading people. We’re not turning on each successive element until we have sufficient volume to achieve a steady-state flow. The advantage here is that from the very first day, we’ll be effectively using our labor, and that’s something you can’t do with a static analysis of rates. Another objective is to be able to really understand the dynamics of staffing plans and the different scenarios, and how product is going to come through. In addition to that is peak planning, and being able to validate the maximum capacity under a specific load.
There’s one more unique benefit that I found in the last simulation project I did. We hired some new management staff, and I was involved in the initial interviews of these people. Interviewing people for any job today in operations is very tough. A lot of people interview very well, but how can you quickly assess their judgment, their ability to pick up on technology in an automated system? Well, we have this three-dimensional simulation, and I can run a loop of simulation and show them how cases accumulate on conveyors. I can ask them, “If you were standing on a platform watching this particular operation, what do you see as happening, what would you say is a cause of some of these elements?”
If I want to buy a simulation package, do I turn to a consultant, buy something off the shelf, or have it created for me?
All of the above. The one I’ve used most recently, in conjunction with a third-party simulation group, is a product called AutoMod [produced by Brooks Automation Inc.]. It’s a configurable material handling simulation tool, and it’s been around quite a long time. You can go to their Web site and find out who’s actually purchased a license for the software itself. You’re typically going to see only very high-volume, highly automated systems, where people have an industrial engineering staff with experienced developers that are constantly making changes, that would be actual end users of a simulation product.
But that’s one end of the spectrum. The other end of the spectrum is clients who might come to us at ESYNC, where they’re looking for a turnkey facility design and warehouse management system project. In the majority of our large-scale automation projects, we strongly advise using some type of material handling simulation to validate the results that we all come up with jointly. My preferred method when working with a client is to select a material handling vendor after the conclusion of our conceptual design. The material handling vendor then presents their detailed design of what they want to do.
Rama Ramaswami is editorial director of O+F.
Although simulation has been around since the 1940s, its widespread use in industry is fairly recent. Until about twenty years ago, the technique of representing the real world through a computer program was largely the domain of academics, and its off-campus applications — bugs and all — were restricted to aerospace and military training.
With the explosion of computing applications in the 1980s, however, all that changed. Production management systems became commercially available, although simulation software was still limited in its capabilities. A breakthrough came in 1983 with the creation of the programming language SLAMII, which allowed the user to model networked, discrete, or flexible events in a single scenario.
Throughout the 1980s, software developers enhanced simulation languages. Models began to achieve quantifiable results, improving workflow and productivity in manufacturing plants.
The 1990s saw the widespread introduction of graphics, templates, animation, verification and testing devices, and other tools. Six years ago Microsoft Corp. introduced a Windows-based simulation program that provided automatic data collection and did not require programming knowledge. Today’s advanced simulation software allows non-technical users to model any system within minutes.
— RR