opt_ FAQ
1. What problems does opt_ solve?
opt_ is designed to solve supply chain planning problems, be it in production planning, distribution planning or a combination of the two. The purpose is to meet one or more business objectives (e.g. maximize demand satisfaction, minimize demand lateness, minimize safety stock violations) while adhering to specific business constraints (e.g. do not exceed defined asset availability, do not exceed defined demand lateness).
2. How does opt_ represent time?
opt_ uses the concept of time buckets instead of having a continuous time representation. Time bucketization can be uniform (all time buckets have the same duration) or telescopic (smaller time buckets in the near future are combined with larger time buckets in the far future). Supply chain planning problems typically have a long horizon of interest, starting from 6 to 9 months and going to 1-2 years when used for Sales & Operations planning support.
3. Is opt_ suited more for planning or scheduling/sequencing problems?
opt_ is more suited for planning. Planning problems are well approximated with linear constraints. The supply chain models are complex, often global, networks, involving both manufacturing and distribution. The business typically has increased reaction time and typically prefers optimal or close to optimal answers that can drive important business decisions.
On the other hand, scheduling/sequencing problems are myopically focusing on a small decomposed part of a supply chain (e.g. a plant, a line). They may have constraints that are not easily approximated as linear. Business is looking for feasible, rather than optimal answers with very fast responsiveness.
4. How does opt_ model a supply chain?
opt_ uses three key concepts:
skus that can be produced or consumed and accumulate material over time, based on defined minimum and maximum limits. Skus can be products, raw materials, components, etc.
processes that can transform one sku to a different sku by utilizing one or more assets. Processes can be used to represent manufacturing activities like the assembly, mix, build, etc, as well as transportation, handling or other activities.
assets that can be consumed over time, based on defined availability. Assets can be machines, machine groups, manpower, transportation or handling capacity, etc. Assets have a predefined availability.
5. How does opt_ generate a plan?
opt_ creates a mathematical representation of the supply chain problem, meaning a series variables and linear constraints. Each business objective is translated into a linear mathematical program (LP) with a specific linear objective. These LPs are solved sequentially, following Goal programming principles, that ensure that the solution of any LP in the sequence is solved to optimality while not deteriorating the solution of any previous LPs solved in the sequence. Note that opt_ creates the LP formulations leveraging the Pyomo optimization framework. Through Pyomo, the LPs is passed to a mathematical solver for solving them to optimality.
6. What mathematical solvers can opt_ use?
The standard version of opt_ uses the HiGHS linear optimization solver. The enterpise version of opt_ can also use IBM ILOG CPLEX, provided that the customer purchases the required license from IBM.
7. Why would someone choose a commercial mathematical solver?
Open source solvers like HiGHS give us the flexibility to develop, test and use opt_ for a variety of supply chain planning problems of low and medium complexity. This is revolutionary, considering that so far, this was only possible with expensive software that only few enterprises could afford. Still, we are aware that for the most demanding problems, open source solvers can be orders of magnitute slower than commercial solvers. For these cases, a commercial solver is needed.
8. How does opt_ compare with a heuristic supply chain solver?
Heuristic supply chain solvers are typically following an order-by-order planning logic, focusing on demand satisfaction as the single objective. Heuristics do not perform well in complex supply chains, with large number of alternates and large degree of material or capacity sharing. They cannot deal with advanced objective interlacing (e.g. safety stock for a group of products is more inportant than forecast for a different group of products) and they cannot understand specific objectives like minimum resource utilization, maximum stock, storage, etc. On the other hand, heuristics are easier to build and maintain, have a lower price point, have a smaller learning curve and are typically faster than optimization based solvers.
9. Where are the data that opt_ uses typically maintained?
For prototyping and testing, data can be directly maintained in excel. In a production environment, data are typically extracted from an ERP and transformed into the opt_ model nomenclature. After the opt_ is generated, selected outputs (e.g. plan orders) are send back to the ERP for execution and to a BI of customer preference for plan analysis and reporting.
10. What kind of objectives does opt_ support?
opt_ supports a variety of objectives: demand related (minimize demand violations and minimize demand lateness from promised or from requested date), inventory related (minimize minimum or maximum stock violations, in qty or periods of cover), asset related (minimize minimum or maximum asset utilization violations) and others. opt_ objectives take filters, so that they pinpoint specific areas of your supply chain. Find more in the manual.
11. How can I visualize the plan that opt_ generates?
opt_ generates exports at two different levels, detailed and summarized. The summarized can be immediately loaded to any Business Inteligence (BI) tool of your choosing to provide the basic reporting dashboards (Asset Utilization, Inventory, Demand). The detailed can also be loaded to be used as a basis for any additional reporting.
When you download opt_ you get an example with Microsoft Power BI that you can use as a basis.
12. Does opt_ have a manual?
Click here for the manual.
13. After going through the manual and checking the unit tests, I still have questions. What shall I do?
You can reach out in our support forum.
14. What versions does opt_ have?
opt_ has two versions:
- The Standard version is immediately downloadable (Win 64) and is free to use. The Standard version leverages HiGHS as the mathematical programming solver. It is provided on an “implement it yourself and operate it yourself” principle. The opt_ support forum can be used to post questions.
- The Enterprise version is available for qualified opportunities on request (email opt -at- optimalcore.com) and in addition to HiGHS also supports IBM ILOG CPLEX (provided the customer purchases the required license from IBM). The Enterprise version is also free to use and it is usually coupled with an implementation and/or a support contract.
15. In what operating system is opt_ available?
Currently, the standard version of opt_ is only available as a Win 64 executable.
The enterprise version is immediately available as a Win 64 executable and can also be compiled in a customer selected Linux variant on request.
16. Do you offer any formal training?
opt_ comes with 40 unit tests that explore all available functionalities. The user manual is also immediately available. OptimalCore is ready to provide additional training services on request (email opt -at- optimalcore.com).
17. Do you offer implementation or support services?
OptimalCore has been providing implementation and support services in the field for +25 years. Please reach out so that we can discuss the level of support you need (email opt -at- optimalcore.com).