Now, it’s a constant daily stream of motorcycles and pickup trucks delivering packages throughout Bangkok. Apparently, their decision worked out well for them since they are still there, even if not all that great for us and the other neighbors.
But that’s logistics in a nutshell, right? Ultimately, it’s the planning, coordinating and transporting of goods from point A to point B in the most efficient, cost-effective manner possible. It’s about people evaluating choices, making decisions and, sometimes, letting technology make the best decisions for them.
In my recent blog A Multi-Purpose Approach to Logistics Modeling, I argued that logistics modeling — especially when used to optimize tactical and strategic decisions — must take a more comprehensive, holistic approach. Isolating and optimizing just a supply chain network creates a suboptimal solution at best.
In the follow-up blog Speed and Flexibility Key to Effective Logistics Modeling, I explained in more detail how a global logistics planning model was developed quickly using the River Logic platform. It incorporated not just the supply chain network but included important manufacturing, and sales data too. The main point was that logistics modeling doesn’t need to take months or years; progress should be measured in days and weeks.
In this blog entry, I expand the discussion to include the eight most critical data requirements needed for advanced logistics modeling. I was able to utilize all of these requirements as part of the global logistics planning model example mentioned above.
Hopefully, by understanding these requirements, it will help you to evaluate and make better logistics decisions in the future.
Basic Requirements
There is a plethora of supply chain-related software; some are used to manage day-to-day decisions, while others utilize mathematical techniques and algorithms to optimize networks, inventory levels, pallets, truck loadings, and other applications.
It should come as no surprise that all similar, basic software require the same types of data for each type of problem solved. For example, if solving a supply chain network optimization problem, at a minimum, a model will need:
- Time period descriptions
- Location information — e.g., name, city, country
- Route descriptions
- Transport mode types and costs
- Product descriptions
- Inventory (source) volumes
- Demand (sink) volumes
This data is necessary and well-understood, so it is not my intent to regurgitate requirements that are easily found elsewhere.
Instead, I will now explain the eight data requirements that differentiate an advanced logistics (e.g., supply chain optimization) modeling software package from a basic software package.
8 Data Requirements to Get More from Logistics Plans and Data Analysis
1. Telescoping Time Periods
Basic software will typically allow for multiple time periods or buckets — e.g., hours, days, weeks or months — but the length is often constant. Sometimes, it’s even pre-defined by the vendor. It might make for simpler data input and solution analysis and reporting, but is that the goal of modeling? (Hint: no, it’s not.)
Advanced software makes no such assumptions. Multiple, telescoping time periods can always be defined to match the problem being modeled. Each model’s period definitions can vary in number and length. For example, a model might require 28 time periods, 12 monthly buckets followed by 16 quarterly buckets.
2. Multi-currency and Exchange Rates
Even if a model’s initial scope is contained within a single country, don’t settle for basic software that does not allow for multiple currencies to be defined and where exchange rates can be actively managed. There is absolutely no reason for any user to have to convert any costs or prices outside a model to fit into a single currency inside the model. I can guarantee it will lead to mistakes later on; especially when many currency exchange rates look similar, as is the case currently with USD, CAD, and EUR.
Advanced software allows for input of cost and price data for each country’s local currency. Costs are then automatically converted in the program to a base currency prior to being read into the optimization matrix.
As a point of reference, the Pulp & Paper Industry model that I built in 2002 (and described in this blog post) has over 70 different currencies defined with exchange rates managed simultaneously in the same model; every country has at least one paper making business.
3. User-Defined Transportation Modes
Most basic software provides choices for different modes of transportation, but some take the easy route and pre-define simple lists like Truck, Rail, Barge, and Container. While this might work OK in some simpler cases, it also might leave the modeler with a dilemma: how to differentiate between different sized truck trailers, which is especially important for truck loading problems?
Advanced software does not restrict the modeler’s choices to pre-defined mode definitions. The modeler defines his own modes at the level of detail necessary for the particular problem being modeled. For example, trucks with 48’ and 52’ trailers can be defined separately, as well as those with refrigeration, flatbeds, and other styles as necessary.
4. Shipment vs. Trip
A shipment is any measurable amount of movement of goods. Using trucks, for example, it can be either by full truckload (FTL) or less-than-truckload (LTL). In modeling terms, however, a shipment is considered different from a trip, which is the completion of a route (or leg) from point A to point B.
When it comes to applying costs, basic software often does not draw a distinction between a shipment and a trip. This can cause problems if, for example, sometimes the freight cost is based on weight, cubes, or some other measure that varies by how full the container is; and sometimes the freight cost is a fixed cost per route/mode combination, regardless if FTL or LTL.
More advanced software allows for costs to be input by shipment, by trip or both. For example, even if a truck trailer has only one pallet, the full cost to ship those goods will be incurred by the trip.
5. Transfer Pricing and Taxes
Transfer pricing is standard in most basic software. Likewise, taxes, which are levied whenever goods are imported into a country and sometimes when exported out of a country, are also usually included. They have a variety of names including tariffs, VAT, and excise tax. When shipping goods within the same company, the tax is generally based on a company’s internal transfer price.
For supply chain network optimization problems, it is important to include all taxes, which can make a tangible difference to the solution. Even in an age when major free trade agreements are being signed by large numbers of countries, usually with some common geographic attribute, they’re still important.
Advanced software always includes for transfer pricing and taxes. The transfer pricing elimination rules will be flexible enough to accommodate any situation — not just a nominal treatment. Taxes not only become part of the model’s objective function, but they are included in a customizable profit-and-loss statement. See the example in Speed and Flexibility Key to Effective Logistics Modeling.
6. Advanced Unit Costing
Most basic software, especially used for supply chain network optimization, have few cost buckets and scarce modeling of details outside of moving freight between nodes in the network. This makes calculating total costs per lane fairly simple. Don’t settle for simple.
Advanced software includes other critical business functions like purchasing, warehousing, even manufacturing and sales, if necessary. Although most logistics planners are not tasked, or even responsible to account for these non-logistics costs, that’s hardly a reason to not want them available if and when needed in the future.
Ideally, as is the case with advanced software, each solution includes both summary and detailed unit costs that breakdown costs into weighted average cost components, from point of origin (e.g., raw material supplier, warehouse, etc.) to point of destination (e.g., warehouse, store location, etc.). These cost components can then be used to create a profit-and-loss statement by product, by country, by business unit, or some other reporting unit.
7. Aggregated constraints
Most basic software does not allow modelers to create their own ad hoc, aggregated constraints. This is a very broadly defined feature, something like the ability to define any higher-level constraint anywhere in the model.
What is a higher-level constraint? An example might be to constrain the total number of refrigerated truck trailers that can visit a particular DC in the same time period. To apply this kind of constraint, every trailer would need an attribute (e.g., refrigerated, non-refrigerated). Then, data structure inside the program lets the modeler input a minimum and/or maximum constraint value. Although not too difficult for any software vendor to pre-program into the package, it can be very tricky to allow the modeler to do this on her own.
Advanced software makes no pre-defined assumptions or limitations. A modeler is free to define aggregated constraints whenever she wants, for whatever purpose. Constraints can even be defined based on other aggregated constraints or ratios between constraints. Any logistics modeling software that cannot provide this functionality is simply not advanced enough to meet today’s business requirements.
8. Integer constraints
Most basic software has some built-in integer constraints. A modeler would not be able to solve a warehouse site location problem, for example, without such constraints. Another example is sole sourcing from a DC, or requiring a certain minimum/maximum number of warehouses to source a product. There are many possibilities.
Advanced software provides functionality to apply these same constraints and many more, wherever desired. This might require a little more training and work than a simple pre-defined form that only requires selecting a value or typing a number, but flexibility must be preferred, especially in cases where tricky, situation-specific integer constraints are necessary and basic software is lacking.
Concluding Remarks
There are many logistics modeling packages on the market today, and a few are becoming quite popular. But, the market has evolved primarily because supply chain optimization modeling software has been allowed to focus on its own silo. Eventually, Integrated Business Planning will win the day. Companies will no longer want their logistics network to be decided based on just a few data points, like warehouse capacity and costs and freight rates. Important influencers, like raw material commodity prices and trade promotion optimization, can have a real impact on logistics in some industries and basic logistics model software does not allow for such considerations — whereas advanced logistics modeling software does.
