Performance Calculation for Order Picking Systems by Analytical Methods and Simulation

“First, the order picking then the stock planning – In the order picking area the most stuff is usually employed. Here the customer service and the logistics quality are decided. The highest costs incur here.” [5] The planning of an order picking system is characterized by the complexity of the system. All attempts to standardize the planning process failed so far. The aim of two research projects at the department for Materials Handling Material Flow Logistics of the Technische Universität München is to develop a holistic planning method for the system planning (rough planning) of order picking systems. Thereby two different methods of resolution are examined. In the first project the performance assessment is made by the simulation of the system. In the second project analytical methods for performance assessment are used. 1. PLANNING OF ORDER PICKING SYSTEMS The order picking is the central function of the warehouse logistics and has significant influence on areas like production and distribution [7]. Despite the trend towards automation, order picking is a costly area in modern logistics systems [13]. This is mainly due to the high personnel section [11]. In order picking systems from a total quantity of items (the assortment) subsets are assorted by a customer order and then sent to the customer [11] [14]. These are the most difficult tasks of the in-house (intra) logistics [4]. This is due to the complexity of order picking systems, because there is a multitude of ways to realize the picking task [11] [2] [10]. Due to the requirements of the order picking, the most efficient mix of the spectrum of order picking technologies has to be selected. It is important to achieve a high delivery quality and simultaneously a high economic efficiency [7]. Often the right solution does not consist only of one specific order picking technology, but from two, three or four different order picking technologies, which can be arranged in an useful combination [1] [2] [10]. Such hybrid or heterogeneous order picking systems allow to adapt the complete system to the specific requirements. There is no standard solution for picking. There is more than the possibility that standard modules, like an automatic small-parts warehouse with picking stations, can be used [2]. It can be deduced, that the planning of order picking systems is the search for a unique solution, which best meets the requirements [1] [10] [7]. 2. REQUIREMENTS FOR AN ORDER PICKING SYSTEM The requirements for an order picking system become higher and higher. The requirements increase due to smaller individual orders, followed by a higher delivery frequency [9] [2] [7] [3]. Furthermore, the growing diversity of items and the high standards of the material availability advance the requirements. The delivery times will be shorter and shorter and thus the cycle times too [1] [2] [7] [3]. Also additional services, such as the labelling of goods for customers, increase the requirement for order picking systems [9]. For successful planning an exact as-is analysis is the precondition [2]. The as-is analysis and the aim planning include the quantification of the requirements and a quantity structure for the planning is created. These data can be divided into assortment structure, article structure, acsess structure, order structure and shipping structure [6][7]. The requirements for an order picking system can be modified after the planning horizon. Within the planning the development has to be assessed. Here it is recommended to work with various scenarios. Thus different business developments can be taken into consideration and the adaptability of an order picking system can be evaluated [12]. 3. ADVANTAGES OF THE USE OF COMPUTERS Often known solutions are used. For example, companies such as Vanderlande or Dematic use planning systems, which propose system variants due to requirements (key parameters). These systems use data of past projects. All attempts to classify the order picking solutions and to standardize the planning process failed so far [2]. Within the system planning, often systems variants are approximately dimensioned and evaluated. Usually the planner uses averaged data, which are extrapolated linearly for the planning. This kind of planning has reached its limits and can not satisfy the actual requirements for the planning of order picking systems [8]. The use of integrated planning software from the analysis of the actual data, by defining scenarios, the dimensions of the order picking system, the determination of performance and the calculation of the investment and costs can shorten the planning process [5]. By the automation of the performance calculation more variants can be considered. This will reduce the risk that beneficial variants are not examined. 4. INNOVATIVE APPROACHES FOR THE SYSTEM PLANNING OF ORDER PICKING SYSTEMS Two research projects at the department for Materials Handling Material Flow Logistics of the Technische Universität München deal with the system planning of order picking systems. In the first research project, which is supported by the AiF (“Arbeitsgemeinschaft industrieller Forschungsvereinigungen”, in engl. consortium of industrial research associations), simulation is used for the planning. The simulation offers the possibility of illustrating the system load by many individual values instead of one average value. The advantage is that the requirements of an order picking system can be considered more precisely. Thus, changes in the system load can be considered over hours, days, weeks or years, as well as changes in the item and order structure. In the second research project, which is supported by DFG (“Deutsche Forschungsgemeinschaft”, in engl. German Research Foundation), analytical methods are used for the performance analysis. The centre of the analytical model is a process description with MTM (Methods-Time Measurement). By using cluster and statistical analysis the parameters for MTM analysis are determined. Through the clustering of orders the requirement structure of the order picking system can be illustrated exactly. 5. IDENTIFICATION OF OUTPUT BY SIMULATION For the simulation of an order-picking-system the orderlines for every pickingregion, the topology of the picking regions and the principle of picking for every picking-region is needed. With these three parts of information the information-, material-flowand organisation-system is fully described for simulation. The statically dimensioning of picking-regions is depending on the assortment (inclusive the minimum ranges of every article) which is provided in this region for the picking process. Principles of picking-systems amongst others are the classical person-to-goods picking with providing articles statically in small parts storage racks or pallet racks without automation technology, the local picking with providing articles statically, also referred to as “zonepicking”, the classical goods-to-person picking in a picking station, which is provided by automatic small-parts warehouse or automatic high rack warehouse, the inverse picking, the person-to-goods picking with manned rack feeders At the department for Materials Handling Material Flow Logistics a library of modules named “BauKom” was composed of the above-mentioned principles under considering the cognition of the AiF-project No. 14601. These modules are used for picking studies by simulation. In each simulation module several parameters can be set, which amongst others are describing the statically dimensions as number of zones, lanes, columns and rows with appropriate lengths of the bin locations as far as the moving strategy of the picker. Subsequent from the simulation module the represented picking-system is build up, which can be connected with other modules if required. By connecting various simulation modules appropriate hybrid and multi-level picking systems can be created. To ensure an continuous flow of planning and the practicability for the planner without consolidated knowledge, the library of modules for simulation was integrated in an planning environment named “PlanKom”. This planning environment is implemented as windows application by the programming language C# of the .NET-Framework. PlanKom performs among the building of key figures also the complete preparation of data, which runs from the creation of customer orders to the transformation of them with the result of picking-orders. The creation of customer-orders is based on an assortment created by the planning environment. In front the planner has to define the horizon of planning and possible variations of the company performance for the purpose of data extrapolation. From this we get a lot of times of inspection. By the planning environment an assortment and customer-orders have to be created for every single time of inspection. To ensure the execution of transformation of the customerto picking-orders a modeling environment is implemented in the planning environment, which can be used for defining the variations of picking systems furthermore referred as models. In these models several parameters have to be set, to ensure the automated build-up of the simulation model. By Knowledge of the topology of the picking system, the allocation of articles to picking-regions and the information first or second level according to an article-oriented picking we have developed a general algorithm which can determine picking-orders from customer-orders for every possible hybrid and multilevel picking-system. The basic time used by simulation for picking-order acceptance and hand-over once in one zone as well as the supposed time for picking of one or some withdrawal units is deposited in every simulation module. But for precise solutions we prefer to calculate the basic time for acceptance and hand-over of one picking-order by MTM(Mean-Time-Measurement)-operation. While simulation is running for every picking-orderline a starting and completion timestamp is taken as well as the part of basis-, walkingand picking-time according to this orderline. Afterwards in evaluation key figures could be build in any detailed level. For example it is possible to apply the item per hour or the workload of the picker to the whole picking-system or to one particular zone in the picking system. But also investigations for specific order-types as customer orders with more than ten orderlines e.g. could be progressed easily. Beside the mean value of a key figure also the minimum and maximum could be collected. With this auxiliary tool comparison of systems as well as optimizations of a specific variant of pickingsystem can be performed whereby the process of planning gains reliability. 6. ANALYTICAL PERFORMANCE CALCULATION As in the first project, an order picking system consists of various modules. Each module represents an order picking process, which are linked depending on the form of organization. The performance of an order picking system can be defined as picks per time unit. The time required to handle an order consists of the basis time, the picking time and the walking time. Thus, the performance (P ) is defined as follows: P ) , ( orders Picks n n = ) , ( ) ( ) ( orders picks way orders basis picks pick picks