Automation of production planning

Integrated order and personnel resource planning with heuristic and optimization methods

As part of a bachelor's thesis at ZHAW, an integrated planning system for automated order and personnel deployment planning was designed and developed. The aim was to enable consistent, transparent, and practical planning under highly dynamic conditions and variable personnel availability, thereby sustainably improving planning quality in SMEs.

The system combines heuristic and optimization-based methods from operations research and enables robust, high-performance, and practical decision support in real time.

Industrial production plant with multiple machines in a factory hall

initial situation

The starting point for the work was empirical data from a sheet metal processing SME in German-speaking Switzerland. The task was to use this data to develop a planning concept that would enable combined order and personnel resource planning.

The underlying database comprised three central elements:

  • Competence matrix for mapping employee qualifications

  • Attendance profiles for describing standard attendance times

  • Work plans that define processing times and the sequence of work stations for each product


While competency matrices and attendance profiles can be created and maintained with little effort, work schedules were considered critical, as they are a necessary prerequisite for any form of operational production planning.

The attendance profiles were based on the contractual working hours and preferences of the employees and were mapped with a granularity of half-days. This assumption was considered practical and allowed for a realistic representation of flexible working hours without unnecessarily increasing the complexity of the model.

In addition, historical order data was used exclusively for the simulation and validation of the system. Since the project was not carried out in direct cooperation with the data-providing company, the database was used to compare the developed concept with a hypothetical reference scenario under identical conditions.

Procedures and methods

  • At the beginning of the work, a targeted literature review was conducted on production planning with a focus on order and personnel deployment planning as well as relevant scheduling and optimization approaches in operations research. Existing approaches were classified and a sound understanding of heuristic and optimization-based methods was established.

    Based on this, the project goal was defined as developing a consistent planning concept for combined order and personnel planning that could be applied under typical SME conditions. The concept should enable dynamic adjustments such as new orders and plan deviations, deliver transparent planning results, and serve as a basis for decision-making in everyday operations.

    Beyond conceptual development, the aim was to implement the developed concept as a prototype planning system and validate it in a simulation-based environment.

    Specifically, the system should enable the determination of reliable delivery dates for newly arriving orders based on simulation, using order and personnel deployment planning created automatically and almost in real time.

  • The design phase was characterized by an open and exploratory search for a suitable solution for combined order and personnel planning under real-time requirements. Although the target vision was clearly defined, existing concepts from literature and practice could not be directly transferred to the present application case.

    An obvious approach was to formulate an optimization model that optimally solves both order and personnel deployment planning. However, due to the high combinatorial complexity of the overall problem, this approach proved impractical for operational use with short response times. Heuristic methods appeared to be more suitable in principle, but there was no known approach that would enable a consistent combination of order and personnel deployment planning in a single step.

    Against this backdrop, a solution principle was developed iteratively that combines the advantages of both worlds. The central idea was to generate order and personnel deployment planning heuristically and then refine it in a targeted manner through optimization. This concept formed the basis for the further methodological and technical development of the planning system.

  • The developed planning concept was implemented as a prototype in Python and embedded in a simulation-based environment. The aim of the implementation was to verify the conceptual assumptions and to comprehensively map the interaction between order and personnel planning under realistic operating conditions.

    The simulation modeled the production process as a dynamic system in which orders are handled as entities by workstations with limited human resources. On this basis, it was possible to systematically analyze the ongoing order situation in the production environment, capacity bottlenecks in terms of personnel and workstations, and key performance indicators such as throughput time, work in progress, and throughput.

  • The performance of the planning system was evaluated using various key figures and compared with a naive reference scenario that reflects typical, deliberately simplified SME planning practices. Both scenarios were based on identical order and resource structures and differed only in the planning and control logic applied.

    The evaluation was carried out using simulation-based experiments, statistical analyses, and graphical evaluations. The results showed consistent significant improvements compared to the reference scenario and confirmed the effectiveness of the developed concept.

Concept development on white paper

solution concept

The solution concept developed describes the central logic of the planning system and the mechanisms with which order and personnel deployment planning can be implemented consistently, dynamically, and in a practical manner. The focus is not on individual algorithms, but on the interplay of planning, simulation, and adaptability in an operational context. The following sections explain the key conceptual building blocks that enable the system to deliver robust and actionable planning results under realistic SME conditions.

  • The core of the solution concept is the simultaneous generation of order and personnel deployment planning. Both planning levels are not created sequentially, but within the same heuristic and form two inseparable sides of the same planning. This ensures that every order plan generated is feasible in terms of personnel right from the start.

    In contrast to traditional approaches, in which personnel deployment planning is carried out downstream or as a corrective measure, this concept takes qualifications, availability, and operational requirements into account during the planning process. Critical conditions are therefore not checked retrospectively, but are an integral part of the planning process. This prevents inconsistent or unsolvable planning situations.

    Building on this consistent initial solution, personnel deployment planning can be further improved in a targeted manner without changing the underlying order sequence. Combined planning thus forms a stable and robust basis for operational decisions, simulation-based delivery date determinations, and transparent assessment of capacities and utilization rates.

  • The solution concept is designed for rolling planning, which is continuously updated and updated. A conscious distinction is made between a fixed planning horizon and a flexible, dynamic planning component.

    The fixed planning horizon serves to ensure operational reliability. Within this period, order and personnel deployment planning are bindingly determined and form the basis for communicating attendance times and operational implementation in the company. Plan changes within this horizon are only permitted in exceptional cases.

    The flexible planning section follows directly on from the fixed horizon and reflects the future order and personnel situation. This area remains deliberately changeable and reacts dynamically to new orders, changed conditions, or deviations from the plan. This enables the system to continuously generate consistent overall planning without jeopardizing operational stability.

    This separation allows short-term commitment and long-term adaptability to be combined. Even in highly dynamic situations, the planning system remains compatible with day-to-day operations while enabling forward-looking assessments of future developments.

  • New orders are not considered in isolation in the solution concept, but are always evaluated in the context of the current and future order and personnel situation. The aim is to realistically assess the impact of an additional order on deadlines, capacities, and utilization before it is bindingly scheduled.

    To this end, the existing planning is used as a starting point and supplemented with the new order. The resulting order and personnel deployment planning is continued on a simulation basis so that the temporal effects along the entire production system become visible. In this way, reliable delivery dates can be determined without jeopardizing existing orders or planning commitments.

    This mechanism makes it possible to dynamically evaluate new orders. At the same time, planning remains stable, as new orders are only incorporated into the operational plan in such a way that they can be realized within the existing capacity and personnel structure.

    Through the simulation-based embedding of new orders, the planning system becomes an operational decision-making tool that not only maps existing processes but also actively supports quotation and prioritization decisions.

  • The solution concept is designed to respond dynamically to changes in the operational environment. New orders, deviations from plans, or changes in staff availability always trigger the same planning mechanism, allowing order and staff deployment planning to be continuously updated.

    Planning is automated and requires little computing time, allowing adjustments to be made almost in real time. This is particularly relevant for operational decision-making processes that require short-term responses, such as when scheduling new orders or dealing with unexpected deviations in the production process.

    Through the consistent use of heuristic methods in combination with targeted optimization, the system remains stable and performs well even with frequent plan changes. This ensures that planning results are not only mathematically consistent, but also remain usable in everyday operations.

Results and added value

The performance of the developed planning system was systematically investigated in simulation-based experiments. A hypothetical reference scenario served as a basis for comparison, which depicts typical, deliberately simplified planning practices of SMEs and differs from the developed concept exclusively in the applied planning and control logic.

Based on identical orders and resources as well as identical production configurations, the results clearly show the operational and structural added value offered by the developed planning system.

    • Every order plan generated can be implemented by personnel right from the start.

    • No subsequent corrections or conflicts between order and personnel deployment planning

    • Avoiding unsolvable planning situations right from the start of the planning process

    Added value – High dimensional stability and operational reliability without manual intervention.

    • Binding personnel deployment planning over a defined planning horizon

    • Simultaneous mapping of future order and personnel situations

    • Clear distinction between planning-related commitments and flexible adjustments

    Added value – Planning becomes reliable for employees and remains adaptable for the company.

    • New orders are evaluated in the context of the existing order and capacity situation.

    • Delivery dates are based on simulation-based planning rather than static assumptions.

    • No jeopardizing of existing orders or operational commitments

    Added value – Well-founded decisions on offers and prioritization without unrealistic promises regarding deadlines.

    • Near real-time updates for new orders or deviations from plan

    • Uniform planning mechanism for all changes

    • Stable planning even with frequent adjustments

    Added value – The planning system remains usable and performs well even under operational pressure.

  • Compared to the hypothetical reference scenario, with identical order, resource, and production structures, the following effects were observed:

    • Significantly higher adherence to deadlines

    • Significant reduction in throughput time and work in progress with constant throughput

    • Improved average employee utilization

    • Statistically significant effects in simulation-based experiments

    Added value – Measurable improvements in key production indicators under identical conditions.

The results show that the combination of consistent heuristic planning, targeted optimization, and simulation-based evaluation can produce robust and practical planning solutions.

The concept developed is particularly suitable for SMEs, where high dynamics, variable staff availability, and real-time operational requirements push traditional planning approaches to their limits.

The project illustrates how clean modeling, appropriate abstraction, and systematic validation can be used to develop consistent and robust decision-making bases that are also compatible with realistic operating conditions.