Rolling Capacity Planning: Planning for Internal Combustion Engines and Electric Mobility Together
Interview with Yenal Ersen (OPTANO GmbH) and Dr. Kostja Siefen (Gurobi)
The transformation of the automotive industry is forcing suppliers to make decisions with significant financial implications. The overall conditions remain uncertain. Yenal Ersen, Head of Business Development at OPTANO, and Dr. Kostja Siefen, Sr. Director of Technical Account Management at Gurobi, discuss how companies can reliably meet the demand for conventional powertrains while simultaneously preparing for the uncertain transformation of all production processes to accommodate a new product structure featuring electric powertrains —using the example of rolling capacity and production planning across multiple plants with a one-year planning horizon. From their respective perspectives, they demonstrate how mathematical optimization in a software application can support suppliers in these planning decisions.
This becomes particularly evident when companies must reliably continue their existing business operations while simultaneously preparing for new production. There are delivery commitments for internal combustion engine vehicles, while the timing and scale of the e-mobility ramp-up remain uncertain. Nevertheless, decision-makers must make decisions regarding capacity, retrofits, and staffing. They cannot wait until demand is certain before taking action. The planning process then reveals very clearly the consequences of this uncertainty.
Turn complexity into value:
To use the many possible courses of action to lay a better foundation for the next decision.
Let’s assume that both businesses share some of the same production facilities and skilled workers. Other resources are product-specific. The supplier must decide when which facilities need to be retooled, what volumes the plants can handle, and where additional shifts are needed. Lead times and customer approvals apply; available capacity cannot be shifted at will.
If he upgrades too early and the ramp-up is delayed, there may be unused capacity. If he waits too long, there may not be enough capacity later on. With OPTANO, you can compare the alternatives and their implications.
The time in between must also be factored into the calculation. A plant does not produce during the changeover. If another plant is to step in, it must be suitable and approved for that purpose. An inventory buffer can only be built up if materials and capacity are available in advance.
The strength of the solver’s mathematical algorithms lies in their ability to consider all factors and their interactions collectively and to balance conflicting objectives within the overall model. Thus, the solver does not simply seek the most favorable production location, but rather a suitable combination of quantities, shifts, inventory levels, and changeover times over the entire period.
You mention OPTANO as planning software and Gurobi as a solver. How do the two work together—and what do planners actually see in practice?
Planners work with the OPTANO application. We translate the client’s specific processes, rules, and conflicting objectives into a mathematical decision-making model and integrate the relevant data. Equally important is usability: Planners should, for example, be able to check a later changeover date and understand what that means for delivery capability, capacity utilization, and costs—without needing to know the mathematical model. They need clear results that they can discuss with the plants.
The Gurobi Optimizer calculates specific planning proposals in the background that satisfy all the conditions defined in the model and best achieve the specified objectives.
Business requirements, such as customer approvals, are stored as rules in the mathematical model. This allows the solver to correctly account for these requirements without needing to understand their business context. OPTANO thus translates the business problem into a model and an application. We provide the technology to efficiently solve the mathematical problem.
No. In our example, the schedule is updated every month. Actual production volumes, inventory levels, and new call-off information are incorporated. The previous month is removed, and a new one is added. This way, the next twelve months are always in view.
Conversions that have already begun or shifts that have been firmly scheduled remain in place as fixed parameters. The application shows which decisions are still pending, what has changed since the last plan, and which plants are affected.
Such changes adjust the model that has already been solved. Decisions that have already been made remain fixed. The solver then optimizes only those planning decisions that can still be changed in practice.
The effort involved in making further plan changes can also be taken into account if it is described in the model. This means that not every small change in demand has to result in a completely different quantity allocation. The solver treats the stability criteria defined in the model either as fixed constraints or as objectives that are balanced against other planning objectives.
Forecasts are always subject to uncertainty. Mathematical optimization helps assess potential deviations and their consequences in advance and develop more robust plans. This creates decision-making capacity and, with it, the ability to make informed trade-offs between stability and flexibility within the available time.
It is important that we do not assume a different decision today for every possible scenario. The only options left open are adjustments that would actually still be possible later on. Under these conditions, the solver calculates the respective consequences. The comparison shows which decisions leave room for flexibility and which ones will prove costly if deviations occur.
We’ll check that first. If the existing systems handle the task well, an additional solution isn’t automatically necessary. A gap can arise if each plant plans independently and the combined effects are only reconciled in spreadsheets afterward. A delayed changeover then triggers multiple rounds of coordination. OPTANO can map these interdependencies within a unified planning process. Data integration and embedding into workflows are tailored to the existing system landscape.
For the calculation, it is crucial that relevant dependencies do not end at plant, system, or planning boundaries. A material allocation that appears favorable for a single plant may cause a bottleneck elsewhere. The solver therefore considers the areas represented in the model and their interactions collectively. This prevents optimizing individual areas at the expense of the overall result.
Even a good demand forecast does not solve this problem on its own: It provides important input data that must first be used to develop a workable plan that takes into account, for example, capacities, approvals, and lead times. It is therefore crucial to define the model boundaries in such a way that the dependencies relevant to the decision are captured.
Then we need to understand why. Is a rule missing? Is an assumption outdated? Or does the proposal highlight a possibility that had been overlooked until now? If, for example, the necessary specialists are not yet qualified by the conversion deadline, this must be taken into account. That’s why we involve users early on and give them the opportunity to review assumptions and compare alternatives. The model complements the expertise of the departments by providing a consistent assessment of all relevant constraints and alternatives.
The solver checks the stored conditions, not the completeness of the operational description. If the description is corrected, the calculation must be rerun. Fast calculations are especially crucial when there is little time left for analysis and adjustments. This applies to both new planning and rescheduling. That is why we are working together with OPTANO to keep the system’s response times as short as possible.
In addition to a plan, mathematical optimization also provides a bound for the best possible result. This makes it possible to estimate how much further improvement is possible without already knowing the optimal plan. This complements the technical evaluation based on KPIs and helps determine whether the achieved solution quality is sufficient or whether it is worth investing additional computation time.
However, this statement refers to the model. It does not represent a savings rate compared to the previous plan. To determine that, the plans must be compared using the same business criteria. These are two different assessments that complement each other.
Exactly. We make comparisons based on the same data and the same delivery requirements. In our example, this includes production, shift, transportation, and changeover costs, as well as inventory levels and delivery capacity. The effort involved in making schedule changes is also factored in.
We then monitor the situation over several planning cycles to see if the operational benefits are confirmed. Historical comparisons are only fair if we use the information that was available at the time. With today’s knowledge, it’s easy to come up with a better plan for yesterday.
No. A sensible starting point would be a defined area where both stores share capacity. The allocation of space to other products must also be taken into account. We need volume estimates, capacities, shift options, inventory levels, and costs, as well as lead times and approvals. Equally important is what has already been decided upon and is binding. The data doesn’t have to be perfect everywhere. However, we need to identify which gaps limit the plan’s validity.
Using what-if analyses, you can determine which data points are particularly important. Does a longer approval time change the recommended changeover date? Does the volume distribution still make sense with a later ramp-up? The solver calculates the results based on the specified assumptions. Your comparison shows where more precise data could influence the decision. However, the calculation does not replace missing information. Rather, it helps to specifically identify the questions that should be clarified before a decision is made.
Similarly, it is possible to assess whether expanding the scope of planning could be economically worthwhile: To what extent would the overall result improve if additional areas or courses of action were included?
A clear description is enough to get started: What capacity do we need for e-mobility, what do we need to continue delivering in the internal combustion engine business, and what decisions lie ahead? Together, we can figure out where our software could help and what the best first step would be. No one needs to bring a finished data package to the table.
We can discuss which alternatives should be analyzed, what conditions apply, and when a result is needed. This way, at our shared booth, we connect the operational task with the analysis behind it. Manufacturing companies often already have planning algorithms in use. We can build on these and work together to identify their limitations and explore potential improvements. New technology must demonstrate its added value. That’s why we take a concrete look at where we can deliver noticeably better results.
Focus on Your Capacity Planning – at the IZB
At the joint OPTANO and Gurobi booth, you can learn how mathematical optimization can support your specific planning task.
IZB 2026, Wolfsburg, Hall 5, Booth 5132
Interview by Kearney with Prof. Dr. Bratzel: „Not a single stone will be left standing“ | Kearney