Demystify Multi-Echelon Optimization
Last month began my 40th year in the Service Parts Planning software space. I have indeed seen it all! From creating the first commercial service parts management solution in 1990 through the growth and innovation of Servigistics, Xelus, and MCA solutions in the 2000’s. Continuing today, PTC has unified the best capabilities of these technologies into a high-value solution poised to capitalize on emerging technologies and data science.
Servigistics represents more than 50 calendar years of experience collaborating with the world’s leading service providers and over $1 Billion in R&D investment. We optimize billions of parts and locations every day! Our legacy of innovation has been validated consistently through earning the highest ranking in EVERY industry analysts survey over the past 35+ years, including this January’s IDC MarketScape.
SOURCE: IDC MarketScape: Worldwide Manufacturing Service Parts Management Applications 2020 Vendor Assessment”, by Aly Pinder, January 2020, IDC #US44801020
Yet, in the face of this reality, if you peruse some vendor websites, you would think we have identical products. They represent a fraction of our time in the market, an even smaller fraction of our R&D investment, and an even smaller fraction still of tier-one service providers, OEMs, Defense Agencies, and Airlines. You could never tell that from what they claim!
The Most Overused, Misunderstood and Diluted Term
Optimization is the most used and abused term in supply chain today. A broad definition allows less sophisticated products to “tick the box” with little to back it up. To some vendors, it is evident that any process using an algorithm to generate a result is termed “optimization”. The reality is that Servigistics stands alone with the only true set of optimization engines developed uniquely for an aftermarket supply chain. You don’t have to take my word for it. The Logistics Management Institute (LMI), at the behest of the US Department of Defense, has certified only one service parts planning tool for use in its constituent agencies. Consultants and OEMs have extensively vetted our math. No one else even comes close to meeting that standard.
The Servigistics Standard
Service supply chains are very unique with numerous critical nuances – huge item and location populations, fluid networks with rebalancing, repairable parts, a preponderance of low-volume demand – to name a few! It takes a considerable commitment in research to develop the algorithms that correctly represent the service supply chain and more investment still in architecture and processing speed to yield results in a reasonable amount of time. Manufacturing supply chains optimize to production with capacity constraints. Instead, service supply chains must meet an increasingly diverse array of goals: equipment uptime, allowable wait time, a variety of Service Level Agreements (SLAs), and good old fill rates. Generic or manufacturing optimization products cannot be adapted to represent the aftermarket.
You Can’t Learn Multi-Echelon Optimization from Wikipedia
It seems every vendor in this space now claims to have Multi-Echelon Optimization (MEO). This MEO facade banks on a misunderstanding of the term and supporting technologies. Servigistics’ leading MEO technology, validated by 3rd parties, is the outcome of a team of PhDs, close alignment with academia (Cornell University, Wharton School at the University of Pennsylvania), and highly discerning clients (US Air Force, US Coast Guard, Volkswagen, HPE, Hitachi Vantara).
Carmen Kubo, Sr. Manager Global Spares Planning at Hitachi Vantara, said it best during her recent LiveWorx presentation,
“I have 30 years of experience in spares planning. Servigistics is by far the best and most powerful planning system out there!”
It would be best if you approached any vendor claiming MEO capabilities with a healthy dose of skepticism. You can’t learn MEO from Wikipedia, and Slideware or Marketecture should be validated. Avoid organizations masking their MEO capability with smoke and mirrors. Trust the data, science, and 3rd party validation. Beware of any vendor presenting their VP of Sales or Chief Marketing Officer as an expert in MEO!
Don’t be fooled by the “Servigistics is too complicated” argument either. MEO is complicated! Truthfully, it is really complicated. That is why it has taken decades of refining, extending the math, and validating the results. The users are insulated from the complicated math, so it is our issue, not theirs. No other vendor has made anywhere near our level of investment. There are no shortcuts, no “partial” MEO. It is simple, you either have it, as we do, or you don’t. The consequences of not having it are severe and can be easily quantified.
The True Optimization Standard
What is multi-echelon optimization? Once and for all, let’s define the standard for true multi-echelon optimization. The only way you can optimize an entire service network is to evaluate the entire service network simultaneously. True optimization algorithms must calculate every part at every location for the entire planning horizon in one run and process in a short time window. Additionally, the algorithms must be able to comprehend the real-world realities such as budget and space constraints while meeting an array of service goals. As soon as you don’t account for everything, by definition, you are sub-optimizing.
With every shortcut comes diluted results. Some seemingly innocuous shortcuts can compromise results entirely! Solutions optimizing one location or echelon at a time, and then adding them up later, ignore the critical interplay between locations in managing demand and supply. This method also significantly increases inventory requirements for a given level of service. You can’t optimize pieces of a supply chain, add them up and call it MEO. These solutions cannot optimize at the product level but instead, add up the constituent parts and present it as optimization. In this model, asset availability and equipment uptime planning is impossible!
Optimization is More Than a Box to Check
How can you validate the optimization algorithms to ensure true multi-echelon optimization? The devil is in the details. We can help by suggesting a few questions to ask that will dig deep enough to expose the level of sophistication of optimization algorithms. Keep in mind, the integrity of these algorithms can mean the difference between meeting your customer expectations or not and minimizing inventory expense with the optimal mix of parts.
- Can you optimize the entire service network simultaneously?
- How many different types of service goals can you solve for?
- Equipment uptime, allowable wait times, service level agreement, target fill rates, others?
- What independent organizations have validated your math?
- Do stocking levels in one location affect stocking levels and predicted service level in another location?
- Can targets be specified by a portion of the supply chain including parts, locations, customers, contracts, and products? Can these groups overlap and how many can be targeted?
- Can a group fill rate be targeted while allowing the system to unequally distribute the fill rates within the group?
Don't Compromise!
The customers for service delivery are ever more demanding. Service parts are the lifeblood of that process. If you compromise in your parts planning solution, you compromise every process it supports. There is great value in taking the time to evaluate a vendor’s capabilities thoroughly. To do so, you need to ask the right questions.
