Tips for Building Supply Chain Models

It is best to create models with not more than 20 – 25 different facilities, and a similar or smaller number of products, vehicles, and routes. Create high level models to simulate supply chain operations on a global or regional scale. And build lower level models of specific sections of a supply chain to simulate those operations in greater detail. If you mix different levels of detail in a single supply chain model, it will increase the model’s complexity and make it harder to understand. It will also cause simulations to run slowly.

As you build your supply chain models, it is important to manage a model’s level of detail and complexity by scaling and adjusting the four supply chain entities appropriately. The discussion that follows will show you how to do this.

Product Icon small2 PRODUCTS – Define a product as a standard shipping case or pallet or multi-unit package of that product. For instance, if Product A is typically shipped in cases of 100 items, then define Product A using the price, weight and volume data for a 100 item shipping case. There is no need to define the cost, weight and volume for an individual item unless you are doing low level modeling of a part of a supply chain in great detail.

It is also useful at times to group similar products together under a single product category or “kit”. For instance, a product labeled “Furniture C” could be a collection of similar furniture items all shipped together in a single 40′ HC cargo container. In that case you would define the Furniture C product using the price, weight; and volume numbers for an entire cargo container loaded with the product. Another product called “Home Hardware” could be a collection of related products (such as nuts, screws, hinges, door knobs, etc.) that are typically shipped together in a single standard package or kit. You would define this product using price, weight and volume numbers for the entire package or kit.

When you open new facilities, you generally want to have a week or 10 days of products on-hand to meet daily demand. As a practice, most companies have a bit more on-hand inventory at new facilities than they actually expect to need, but this extra amount acts as a buffer against uncertainty. After all the analysis is done, a new facility is still a new facility with no history, so there is uncertainty. On-hand amounts can always be adjusted later in simulations for following planning cycles (scroll down to TIMELINE to see more about sales and operations planning, or S&OP, cycles).

There must be at least two days of product on-hand to meet demand at any facility (new or existing). One day is needed for new supplies to reach a facility, and at least one day’s supply of product will be consumed in the “Day 0” calculations that start all simulations (scroll down to “DAY 0″ CALCULATIONS AND ADJUSTMENTS” to see more about these calculations).

Learn more about how to reduce inventory and operating costs in the section titled “Cutting Inventory and Operating Costs

Facility Icon small2 FACILITIES – Depending on the level of detail of your model, you can have one facility stand for a group of facilities in the same geographical area, or there can be a different facility created in the model for each actual facility. For instance, if you are modeling a retail supply chain on a national or regional level, it works best to simply create one store in each city and have that one store represent all the individual stores in the city. If you are modeling a local supply chain in one city then you can create facilities in the model for all the stores in that city.

When one facility stands for a group of facilities, define that facility as having the combined storage capacities, production rates, demands, and costs of all the facilities it represents.

Assign on-hand inventory to each facility as it should be at the start of your simulation. Then as the simulation plays out, you will see how on-hand inventory changes. Assign production per day to facilities that make products (such as factories), and assign demand per day to facilities that consume or sell products (such as stores). Do not assign demand to facilities where products are only stored for delivery elsewhere (such as warehouses) because the on-hand inventory will be reduced as vehicles take products from those facilities and deliver them elsewhere. Also see “DAY 0″ CALCULATIONS AND ADJUSTMENTS section below to understand how on-hand inventory at some facilities may be adjusted at the start of a simulation.

Many individual costs associated with facilities such as: inventory carrying costs; cost of utilities, cost of insurance, cost of maintenance, cost of labor, etc. are not recorded as separate costs. Instead, they are included in the two main aggregate costs assigned to every facility. Those two main costs are Daily Rent Cost and Daily Operating Cost. Other costs such as inventory carrying costs are themselves aggregate numbers composed of costs such as: cost of capital invested in inventory; cost of rent; cost of utilities; cost of taxes; cost of handling etc. Therefore, make estimates of inventory carrying cost and other relevant costs when you calculate the Daily Operating Cost for a facility. In different 30-day periods those costs may vary so, run different simulations with those variations as you explore their effects in different 30-day periods.

At factories, daily production rates can vary. So therefore in 30-day simulations experiment with different production levels and observe what results occur. Set safety stock levels based on simulation results that reflect your best forecast of what daily production rates will actually be.

If you want to focus your model on a supply chain that begins with a certain facility such as a factory or warehouse, you do not need to show the supplier facilities and routes that deliver products to that beginning facility. Instead, you show those incoming products as being produced at that beginning facility. Show the products as being produced at a daily rate equal to the average daily delivery of those products over a 30 day period. Relevant costs associated with the delivery or procurement of these products should be added to the daily operating cost for that facility.

The same is also true for a facility where a supply chain model ends. If you are modeling the supply chain of a manufacturer or a distributor, then the supply chain model may well end with a warehouse. In that case, show product demand at the warehouse as equal to deliveries made to retailers who get their products from that warehouse.

Transport Icon small VEHICLES – Depending on the level of detail of your model, one vehicle may represent many individual vehicles as a single group. For instance, if a fleet of 10 trucks is required to deliver products to a given facility, then simply create one vehicle that represents 10 trucks. For more detailed and lower level models you can create separate vehicles in the model for each truck.

When one vehicle stands for a group of vehicles, define that vehicle as having the combined cargo volume and weight capacities and costs of all the vehicles it represents. This vehicle will then have one route that all the represented vehicles will follow.

Assign speed to a vehicle that represents its average speed over the route it travels. That means if the route has heavy traffic or bad roads the vehicle will go slower than its top speed. Also factor in the time the vehicle spends parked at stops on its route where it delivers or picks up products.

For vehicles such as trains, airplanes and ships you are usually working with shipping containers that are carried by these vehicles. For instance, you define a vehicle used to deliver products across an ocean as a ship, and the cargo capacity is determined by the number of shipping containers you use on that ship. If you use one standard 40 ft. shipping container then the volume and weight of products you can ship are determined by the capacity of that container. If you need more capacity then add more containers. For instance, you can define a vehicle as being ten 40 ft. shipping containers, and this vehicle would have ten times the capacity and cost of a single 40 ft. container. Cargo carriers will tell you the cost of different sizes of containers so use that to calculate the operating cost of the vehicle you create.

Roll-on/Roll-off (RORO) is used for shipping wheeled vehicles such as cars and trucks. If you want to create a supply chain model using RORO then define the container size to be the same as the size of one or more wheeled vehicles. If you are using a RORO vessel and want to ship 100 cars and each car is 9 cubic meters in size, then define a container of 900 cubic meters and it will accommodate those cars. Cargo carriers using RORO will tell you shipping costs per vehicle so use that to calculate the operating cost for this 900 cubic meter container.

When the vehicle is a ship you may want to consider special ways to model some delivery schedules. For instance: if the route a ship travels has a round trip time of two or three weeks at average ship speeds (about 30 km/hr) and yet you want to show shipments being made and received once every week, then increase speed of the ship vehicle so the round trip time is reduced to one week and set the delay between departures to be zero. Now the model will record a stream of shipments leaving each week from an originating facility and arriving each week at a destination facility. When you do this, be sure to set the vehicle operating cost to half of the normal cost to accurately model the cost of one-way shipments instead of round-trip costs.

Use this technique when modeling one-way shipping containers to move products from one facility to another on a regular schedule without worrying about when a particular container (or vehicle) returns to its facility of origin. You would typically use this technique when modeling container shipments by ship, rail, airplane or truck. You would not do this when modeling the use of company owned delivery trucks, or vehicles where full round trip costs are paid for.

At times it may be useful to make a one-time shipment of products from a facility. Do this by adding a new vehicle that can carry the required volume and weight, and set its delay between departures to be 1,000 hours. The vehicle will depart once at the start of the simulation and deliver products to facilities specified on its route.Then it will return to its starting facility and wait 1,000 hours before departing again. That means it won’t depart again until after a 30-day simulation period is over.

See the “DAY 0″ CALCULATIONS AND ADJUSTMENTS section below to understand how total vehicle operating costs are adjusted at the start of a simulation.

Route Icon small ROUTES – In a supply chain model built at a global or national level, the exact path taken by a route does not have to be precisely accurate. As you create lower level models at the region or city level, it is more important to have a route follow more closely to the exact route taken in the real world. Zoom in on the map when creating the route and adjust it as needed to get this extra accuracy.

When you model the route of a ship or barge that travels up and down a river or canal, you start with the route shown as a straight line drawn between the originating facility and the destination facility. You will see small white globes placed at intervals along the route line. Drag and drop those small white globes as needed to make the route line better follow the real path of a river or canal.

When you create a railroad route it will follow the nearest road because usually the roads and railroads run close enough together. However there are times when that is not the case and you want a more accurate path for a railroad route. In that case, define the vehicle as a ship or airplane and the route will draw as a straight line between the origin and destination facilities. Then you can drag and drop the white globes along the route line to adjust the route to follow the actual rail lines more closely. When you save the route, then go back and reset the vehicle type to a train and adjust cost and operating numbers as needed.

You can replenish product safety stock at facilities by setting product drop quantities on routes that service those facilities to be slightly higher than product demand. This practice of replenishing safety stock produces a saw-tooth pattern in the on-hand inventory graphs for products at facilities. That is because on-hand inventory starts to build up and then gets drawn down again when a missed delivery occurs and the accumulated safety stock is consumed to cover that missed delivery (learn more about interpreting patterns produced by on-hand inventory graphs in “Analyzing Simulation Data“).

A good way to estimate the amount of product to deliver to a facility on a route is to use the formula for Economic Order Quantity (EOQ). This formula is a way to estimate the best amount of a product to order and when to place the order given the demand for a product at a facility. Set product delivery amount (drop qty) at a facility to the EOQ amount for the product at that facility (you can assume the EOQ amount is what the facility ordered). Divide EOQ by daily demand for the product at that facility to estimate how often deliveries should be made. There is a complete explanation and an example of how to use data available in SCM Globe to calculate this formula – see the section “Lowering Inventory and Operating Costs” — scroll down to the heading “Use Economic Order Quantity (EOQ) for Delivery Amounts and Frequencies”

Clock logo small TIMELINE – The default time unit used by SCM Globe is the hour. Hours are aggregated into days for display in the simulations. If you want to model and simulate periods longer than 30 – 60 days, you can scale the timeline. For instance, in your model one day could represent two days, or one week, or one month instead of just a day. To use this technique, adjust your model so that daily operating costs become two-day, or weekly or monthly costs. Set production and demand rates equal to two-day, weekly or monthly rates as well. And vehicle speeds also increase by a factor equal to the number of days represented by a single day in your model. For example, if you want one day to equal a week (7 days), then increase vehicle speeds by a factor of 7.

The reverse will also work. If you want to make one day of a simulation equal half a day or 6 hours or one hour, then decrease demand, costs and speeds accordingly. You cannot make simulations that measure time units below one hour unless you do this reverse scaling of the time line. If you want to simulate delivery routes where the round trip time is less than one hour, then you need to employ this reverse scaling of the time line.

Simulations are typically focused on 30-day periods, not 6-month or annual periods. This is because most supply chain professional organizations recommend a 30-day sales and operations planning (S&OP) cycle. In the S&OP process companies forecast product demand for a 30-day period because 30-day forecasts are inherently more accurate than 90-day or 1-year forecasts. High rates of change and unpredictable events in the global economy make it difficult to create accurate forecasts for periods beyond 30 days. This means companies forecast and plan their way through an uncertain world by using 30-day forecast windows.

You may wish to simulate an entire year by using a sequence of 30-day simulations that cover a 12 month period. Do that by setting the production and demand levels in each monthly simulation to equal what you expect those levels to be at that time of year. Set daily demand higher in times of expected higher demand and set daily demand lower in times of expected lower demand. To simulate unexpected disruptions at given times in a year or month, pick a month and run simulations with different levels of demand or production or product on-hand amounts at certain facilities and see what happens.

For instance, if you are simulating a supply chain to handle gardening products, you know demand will be higher in summer months and lower in winter months. To see how such a supply chain performs at different times of the year, set product demand at facilities to be what you expect in a typical summer month. To see how the supply performs in winter months, set product demand at facilities to expected winter levels. Or set product demand levels higher than expected for a winter month and see what happens. Adjustments each month to entities in the supply chain (products, facilities, vehicles and routes) are needed to make the supply chain run well in a high demand period versus a low demand period.


Once you create or update a supply chain model, you bring in the SCM Globe simulation engine to drive that model and see how well it performs. There are some initial calculations made by the simulation engine that affect beginning amounts of on-hand inventory for some products at some facilities. These initial calculations create the “Day 0” starting point for the supply chain at the beginning of a simulation:

  • On Day 0 the facilities that have daily demand for a product have that daily demand amount subtracted from their on-hand inventory of that product.
  • Vehicles leave their facilities on Day 0 with their delivery quantities which are subtracted from on-hand inventory amounts at the facility.
  • Also if there are routes originating at a facility where products are picked up from other facilities and returned to the that facility, then the amount(s) of picked up product(s) are added to on-hand inventory at that facility.

These adjustments on Day 0 reflect the fact that, in most cases, supply chains are on-going operations already in motion. This means the model must make a running start into the simulation and not simply act as if all operations began on day one of the simulation. These running start adjustments do not change the overall supply chain patterns, trends and performance levels that are shown in the simulations. And most of the time in most supply chain models it is not worth worrying about these things.

However, in some cases you may wish to compensate for these Day 0 calculations. You can do this by adding an extra day’s amount of on-hand inventory for product demand at affected facilities. And you can subtract pickup amount(s) from on-hand amounts of product(s) at affected facilities.

In addition, Day 0 calculations also add a fixed cost component to vehicle operating costs. This fixed cost component reflects costs such as maintenance and repairs, insurance, taxes and other overhead. It is simply calculated as a percentage of variable costs. Variable Cost is cost per kilometer times the number of kilometers traveled each day.

Fixed costs for vehicles are front loaded at the start of each simulation. This means that over a short simulation time span such as 1 – 5 days it amounts to approximately 50 % of total vehicle costs. Over periods such as 7 – 15 days, fixed cost drops to about 30 % of total vehicle cost. And by 25 – 30+ days, fixed cost drops to around 25 % of total cost. This method of calculating costs is conservative and provides more accurate estimates for transportation costs in a supply chain than if we were to use variable costs alone.


For modeling purposes we pick supply chain boundaries that fit our needs. For instance we may decide that a supply chain starts with a certain factory even though that factory is supplied with raw material or component parts by other factories. But we can just model the products produced by the factory and not worry about what was supplied to the factory to make those products. We can capture the cost of those inputs in the daily operating cost of the factory if we wish.

If we are creating a more detailed model of a part of a supply chain, we may start with a warehouse that receives products from sources outside the model. In that case, just define a daily production rate for those products to the warehouse. This models the daily delivery of those products to the warehouse without having to show where they came from.

Also when modeling a portion of a supply chain we may be interested only in the part that connects factories and warehouses and not want to show the stores where products are ultimately delivered. We can handle this by placing demand for products on a warehouse. In this way we model the daily consumption of products without having to show where they were delivered.

When the model does include the stores that a warehouse makes deliveries to, do not define any daily demand for products at the warehouse. That demand will be reflected by the daily deliveries of product made by vehicles on routes starting at that warehouse. Those deliveries will reduce on-hand amounts at the warehouse and if demand numbers for those products were entered for the warehouse, it would result in a double counting of the demand.


Advanced supply chain modeling, simulation, optimization, and risk assessment services are available through our consulting partners. They can work with you online or in person. Find out more by inquiring at or contacting Michael Hugos at


Copyright © 2016 by SCM Globe Corp.