You are on page 1of 10

Automated Strength Analysis for Propeller Blades

Sebastian Daniel Bade, Gemanischer Lloyd, Hamburg/Germany, sebastian-daniel.bade@glgroup.com Andreas Junglewitz, Gemanischer Lloyd, Hamburg/Germany, andreas.junglewitz@gl-group.com Abstract In this paper the modular tool chain is illustrated that enables the Germanischer Lloyd (GL) to cope with heterogeneous design representations handed in for approval by propeller manufacturers. It is applied to perform an automated calculation of the design as precondition for a quick plan appraisal. Various professional engineering tools of the presented approach are interlinked. The Friendship Framework is used to create, examine and mend a parametric geometry representation of the designs. A lifting surface code is applied to determine the hydrodynamic loads for the propeller and ANSYS is used for the calculation of blade stresses. The work-flow and the coupling of the individual software tools are controlled by a set of scripts while the interactive user interface for this controller is again supplied by the Friendship Framework. The pros and cons of this modular approach will be discussed. 1. Introduction The ships screw propeller is one of the most essential component related to ships safety. Therefore the design of a propeller has to undergo a plan appraisal by a classification society. Within this plan appraisal a check of the propeller blade strength is performed. This check is based on simple formula as well as on FEA calculations. The simple formula is derived from beam theory. It can only be applied to old fashioned blade designs with low skew back. The skew back characterizes the backward sweep of the blade centreline from the direction of rotation (figure 1). The contour of the blade is not radially symmetrical about blade center axis. This geometry causes a twisting of the loaded blade and the resulting stress is not covered by the simple formulas. Propeller designs with a large skew back require a stress calculation taking the real 3D blade geometry and the hydrodynamic loads into account. The flow field behind the ships hull is inhomogeneous. The propeller operates in the wake behind the ship that varies in direction and speed. The propeller blades are rotating around the centre line of the propeller and experience a variation of the onset flow due to the wake distribution in the area of the propeller disk. The inhomogeneous inflow causes a variation of the load in the magnitude as well as in radial and chord wise distribution. To evaluate the design of a propeller it is necessary to identify the circumferential positions of the blade in the wake field that cause extreme stresses. The wake field behind the ship is normally measured in a towing tank of a ship model basin and scaled to the real ship. Such a wake field has to be applied in the calculation of the 3D hydrodynamic loading of the propeller. Currently the Germanischer Lloyd (GL) applies a very reliable vortex lattice method to calculate the hydrodynamic blade loads in the various wake field positions. This method requires information about the blade outer contour, the distribution of thickness and other data derived from the geometry of the design. From the calculation results the most critical positions of the blade in the wake field as well as the associated load distributions have to be selected based on pre-defined criteria such as bending moment and spindle moment acting on the blade. Additional criteria for the plan appraisal of the design are the loads created when the propeller is acting in ice. These loads are modelled by applying surface loads on certain regions of the propeller blade as specified in the rules. While a very detailed geometric representation of the propeller is of minor importance when calculating the hydrodynamic loads more facts are needed for the calculation of the strength. Such details include the hub or the transition between hub and blade. The stresses are calculated using finite element analyses (FEA).

369

Fig.1: Skew angle and other propeller parameters The FEA is used more frequently nowadays due to three reasons. At first the number of unconventional propellers, e.g. with a skew angle in excess of 25, has grown considerably. Secondly, newly introduced ice rules require a screw propeller blade FEA and FEA has become simple to be applied on a conventional Windows or Linux desktop computer. Nevertheless, the geometry treatment is a critical point in the preparation of a FEA calculation. The entire ships screw propeller is a solid object represented by a very complex 3D geometry. Starting from the set of global and local parameters used for the hydrodynamic calculation a 3D surface representation of the propeller has to be created including all the details that are required for the detailed stress analysis. The essential task is the creation of a closed surface representation that can be seamlessly processed in the typical FEA software tools. The components of this integration include the generation of a mesh for the FE-model and the mapping of hydrodynamic and ice loads as well as boundary conditions into the mesh. The challenges in this process are the variety of geometry definitions used by the different designers, propeller makers and model basins. The community lacks a standard for the propeller geometry. In this paper an automated work-flow to evaluate the propeller design at a classification society is introduced. For this approach a lean set of essential parameters characterizing the design from hydrodynamic as well as from strength point of view is used to feed the process. 2. The parametric definition of the propeller design In the past the data defining a propeller design and the geometry details were created and communicated on paper in the form of drawings, offset tables and calculation sheets. Nowadays, nearly the complete propeller design process is software based and generally the digital information produced in the process is suitable to support the appraisal of the design by a classification society. The software systems used by the designer support the evaluation of the hydrodynamic performance of the design. It is also possible to examine the strength characteristics or the manufacturing process of the new design. One would expect that such information can be directly handed into the process of design appraisal at a classification society. Unfortunately there are several potential conflicts hindering a seamless integration of the data. A detailed CAD representation of the design may expose much of the manufacturer's know how and understandably the suppliers tend to protect their intellectual property. Another typical problem with CAD data is the transfer of data into other systems which yields corrupt or not entirely complete data models. A classification society or other service providers may not have the same CAD system available. Many times the imported CAD representation contains defects due to differences in the modelling approach.

370

From the standpoint of a classification society using a detailed CAD file of the ship propeller does not render a comprehensive representation of the design in the appraisal process. The classic rule based approval approach is based on some characteristic parameters that are difficult to collect form a 3D CAD model. The parameter values defining the propeller design together with the design drawings are considered to be the mandatory set of information used by the classification society. It can be checked and stamped. Such values can easily be cross checked with the real propeller. Additionally the scope of the design appraisal is entirely documented by the parameter representation and the drawings of the design. In naval architecture it is common practise to eliminate the need to deal with the detailed definition of the 3D blade surface by using a collection of propeller parameters that ultimately represent the surface. These parameters include both global and detailed parameters. These parameters define the overall size and shape of the propeller. Other parameters render 2D blade section and how the wing shape actually moves in the fluid. Further values typically identify where the 2D blade sections are located in space, as well as considerations for clearance, strength and manufacture. In the process for the automated strength analyses of propeller designs described here the geometry representing parameters are stored in a file-format called PFF (Propeller File Format), Schulze (1998). This format is currently used by makers to transfer their geometries to Germanischer Lloyd in order to speed up the approval process Brunk (1998). A characteristic section of this format is presented in Fig. 2. The PFF format contains the information necessary to properly reproduce the hydrodynamic characteristics of the propeller. The elementary strength characteristics of the propeller design are rendered fairly well by a propeller geometry created using these parameters. Details of design elements such as the leading edge or the propeller tip or the anti singing edge are not subject of the appraisal as they usually dont contribute significantly to the overall strength. These details are not part of the PFF file. The type of fillet used to smoothly connect the propeller blade to the hub can be extracted from the design drawings and common parametric definitions. These assumptions are sufficiently precise to allow an overall judgement of the global stresses experienced. The hub design is also characterized in the drawings handed in.
* PropDiameter / HubDiameter / Scale / ExpArRatio / 3600.00 980.000 1.00000 0.460000 * MomentInert / ShaftPower / 3410.00 1800.00 * No.Blades / Material / No.Radii / No.CordPart / 4 3 12 13 * r/R / r / CordLength / 0.272 490.000 584.000 * Pitch / DistLeaEdge / MaxCamber / MaxThick / 3546.000 292.000 19.300 126.900 * Station / Dist.SucS / Dist.PressS / 0.025000 22.600000 -16.500000 0.050000 32.200000 -21.700000 0.100000 45.500000 -28.200000 0.200000 63.000000 -36.000000 0.300000 74.000000 -40.800000 0.400000 80.500000 -43.400000 0.500000 82.700000 -44.200000 0.600000 80.700000 -43.000000 0.700000 73.200000 -39.000000 0.800000 58.800000 -31.700000 0.900000 35.000000 -21.200000 0.950000 20.100000 -13.500000 1.000000 5.000000 -5.000000

BladeMass / 0.

Orientation / 1

Rake 0.000

Fig.2: Sample section from PFF propeller parameters

371

In summary the parameters defined in the PFF render an adequate model representation of the propeller and contain the details necessary for a well-founded global strength analysis. Consequently this kind of geometry definition is used as solution for the problem of the large variety of different user defined formats and CAD models handed in for the appraisal process. Also the ASCII data can easily be checked against the officially stamped documents such as drawings and offset tables. 3. Modular automation work-flow In this section the general work-flow for the automated strength analyses of ship propellers is presented. The focus is on the basic procedures involved as well as on the elements the process is compiled of. The central idea is the application of a modular tool chain that can be executed without further manual user interaction. This way it is possible to enhance the whole work-flow by replacing single modules. These modules can be developed and tested individually. The Friendship Framework is used interactively as tool to control the complete work flow. The automation of the strength analysis for propeller blades is composed of several specialized programs and scripts. Two different systems are in charge of the general flow control. The Friendship Framework provides the user interface for the creation and manipulation of the propeller geometry and the PFF parameter file. It also acts as the user interface for the calculations. The execution of the individual calculation steps and the transformation of data are controlled by Python scripts. This set up has the advantage that the whole calculation work-flow is separated from user interaction and can be developed independently. Given a correct PFF file, a wake-field file and the parameters for the calculation processes the whole strength analysis of the propeller blade can be run without user interaction. On the other hand the user interaction can be developed without running the calculations.

Fig. 3: The modular system of the automated propeller appraisal work-flow The overall activity consists of the preparation of the PFF file containing the parameters defining the propeller. A next step is the generation of a proper 3D surface geometry of the entire propeller for the

372

preparation of the tetrahedral finite elements mesh. In parallel the PFF file and a given wake field is used for the calculation of the hydrodynamic loads. From all resulting loads relevant cases are selected for further processing in the FEA. The tetrahedral mesh, the load cases and the boundary conditions are used to execute a FEA calculation. The calculation results are transformed into figures displaying the stresses in the propeller blade. The final step in the appraisal process of the propeller is the review of result by the classification engineer. The entire process is illustrated in Fig. 3. 3.1 Creation, checking and editing of PFF files Starting point for the automated strength analysis of ship propellers is the PFF file. There are basically three use-cases to be considered: checking and editing of existing PFF files and creating new PFF files. In many cases a proper PFF file representing the propeller can be provided by the propeller manufacturer submitting the design. Even though the PFF file is a human readable format the engineer at the classification society needs a tool to visually check and interactively manipulate the parameters. Another regular situation is that there is only a design drawing and parameter sheets handed in for appraisal. In such a case the starting point of the process is to load a general PFFtemplate and to interactively adapt the PFF parameters to the information available. There are several ways to present the parameters of the design. A general view is a diagram of the 2D section used for the design, Fig. 4 (bottom). Such a diagram can be very useful to check the sectional data. Typing errors are easily detected on the figure as the curve representing the section has discontinuities.

Fig. 4: PFF editing in the Friendship Framework Another useful illustration shows the curves of the radial distribution of pitch, chord length, rake, skew, maximum thickness and maximum camber, Fig. 4 (bottom left). These curves characterize the propeller blade. They are typically used by the designer and for the experienced engineer to create an idea of the hydrodynamic properties of the propeller, Harries and Kther (1997).

373

The PFF file is imported to the Friendship Framework and transferred into a parameter based representation of curves and surfaces. Those can be presented in different views. Several diagrams can be arranged in a user defined set up storable in the system configuration for later reuse. Generally it is possible to use the Friendship Framework to interactively manipulate the parameters of the propeller design by picking a point on one of the curves and change its value by pushing it to a new location. The changes to the parameters can be stored in the PFF format for further processing. 3.2. Creation of the propeller surface geometry In many cases of the failure of an automated CAE process the problem is caused by a faulty geometry representation. From a general point of view the reason is the misinterpretation of the different aspects of 3D geometry. There is the computer generated image representing the illusion of a real worlds object to the viewer. On the other side there is the 3D object modelled in such a way that it reflects the objects physical properties in a calculation process such as CFD or FEA. For each calculation process a different model of reality has to be created. Usually these models are incompatible to each other. For this reason each CAE calculation typically requires an individual model representation. The task here is to create the complex 3D geometry of the propeller in such a way that it can be reliably transformed into a mesh representation applicable in the FEA. The central challenge is that this 3D geometry is composed of boundary surfaces without any gaps. In the automation process described here the surface model of the propeller is constructed using Friendship Framework. Its capability to handle parametric geometry entities and its feature definition language facilitate the tools necessary to create a closed 3D surface boundary representation of the propeller. The creation of the suction side and the pressure side surfaces of the blade is a straightforward skinning of the sectional data from the PFF file, Fig. 5. These two surfaces depict the topological anchor objects for the entire propeller. In order to create a tetrahedral mesh for the FEA calculation, a closed 3D shape is necessary. In addition to the blade surfaces, a hub has to be modelled. This hub has to be smoothly connected to the blade using fillets so that flow and strength criteria are fulfilled. Additional fillets are needed at nose, tail and tip of the blade to close the shape.

Fig. 5: The basic propeller definition contained in the PFF

Fig. 6: Topology of the propeller surfaces

374

The main challenge is to identify a topological set up for the network of surfaces that form the entire propeller including the blade, the hub and the fillets in such a way that the manifold designs coming in for appraisal can be handled similarly. The goal is to reuse this topological set up. In the initial step presented here this topological set up is created for the traditional fixed pitch propeller. The hub is represented by a twisted slice of a cone. The transition between blade and hub is the root fillet and in the future hub designs are developed to better synthesize other propeller design variants. For the current state of development the resulting topological closed shape is shown in Fig. 6. 3.3 Calculation of hydrodynamic loads The hydrodynamic loads are calculated using the vortex lattice flow code VORTEX provided by the SVA, Potsdam. This code takes the PFF file to generate the geometric model of the propeller blade for the calculation. Additionally a wake field and the operational parameters such as propeller rotational speed, ship speed have to be appended. The VORTEX code calculates the loading of the propeller blade in a series of incremental angular positions. For each angular position of the blade the local wake field is applied. Each individual result is created containing the hydrodynamic force vectors at the positions of the calculation grid representing the propeller blade. In order to determine the critical load cases and their angular position the extreme values of bending moment at blade root positions and extreme values of the spindle moment around the blade axis are calculated. The identification of the relevant load cases is performed by a set of Python scripts integrated into the automated process. The result of this procedure is a set of load cases prepared for the FEA calculations. Some propellers are intended for the operation in ice covered waters. They have to undergo additional strength analysis in the appraisal process. The load cases to be applied depend on the type of propeller and are defined in Rules for Construction of Sea going Vessels, GL (2011). Certain areas of the blade have to be loaded with a static surface pressure, which is determined taking the ice class and individual geometrical and performance data into account. Taking the geometric representation of the blade modelled using the Friendship Framework the ice load cases can easily be generated by a feature. Samples of these areas are shown in Fig. 7.

Fig. 7: Samples of blade area with ice load 3.4. Creation of a mesh for FEA Typical screw propellers are represented by a fairly complex 3D surface geometry. This is also true for the geometric representation discussed here for the automated strength analysis. To simplify the mesh generations the tetrahedral elements are chosen for the FEA calculations. The tetrahedral elements mesh can automatically be generated using the tools provided together with the FEA software system. Precondition for the automatic mesh generation is usually a closed boundary representation of the object. The Friendship Framework is used to generate the 3D propeller geometry from its parameter based representation. By applying the topological set up sketched above, the set of 3D surfaces representing

375

the propeller is closed by default and forms a closed shape. The trouble here is that the Friendship Framework currently provides no interface for the export of closed shell boundary representation. To export the geometry to a FEA system, the IGES surface representation could be utilized but this transfer-format does not carry the topological net of surfaces properly. Usually receiving systems are not capable to reconstruct the closed shape of the propeller when transferred in the IGES format. Another option is to transfer the propeller as tessellated geometry. This technique is commonly used for complex geometries. The transferred data consists of a large number of triangles that represent the original shape. The typical format is called STL. This procedure is less error-prone at the price of a lower geometric accuracy. At the current stage of development the STL transfer is the method used in the automation procedure discussed in this paper. With the availability of a better method a new transfer module will be plugged into the modular process. A public-domain tetrahedral mesh generation tool is used that reliably creates the mesh from a tessellated geometry representation. This tool is netgen, Schrderl (2003). This way the tetrahedral mesh is created independently from the target FEA system. An extensive control for the density distribution of the net and the size of cells is possible. Using the tessellated propeller geometry representation, a tetrahedral mesh is generated, Fig. 8, in netgen.

Fig. 8: A propeller blade with tetrahedral mesh 3.5 FEA calculation of blade root stresses Prerequisites for the FEA calculation of the blade stresses is a tetrahedral finite element mesh, a file defining the loads and the boundary conditions. In the procedure presented here, the commercial FEA-software ANSYS is used for the calculation of the nodal displacements and the resulting stresses. The automation of the FEA workflow is done using APDL macros. APDL stands for ANSYS Parametric Design Language and is a commonly used scripting language for the automation of common and reoccurring tasks in the FEA calculation process. As a first step the hydrodynamic loads have to be mapped onto the nodes of the tetrahedral FE mesh. This is performed by an algorithm that transforms the hydrodynamic loads acting at specific positions of the hydro-grid representing the blade into forces acting at the nodes representing a similar position in the tetrahedral mesh. This procedure can be applied for a global strength analysis. The same mapping procedure is used to apply the fixed boundary condition to the surface nodes of the hub segment. This procedure is reused to apply the forces resulting from ice loads to the blade nodes.

376

With the availability of a mesh, the loads and the boundary conditions a FEA can be started. It is controlled by an APDL script that sets all necessary parameters and triggers the calculation. All load cases are processed in batch mode. The resulting nodal displacements have to be transformed into a material stresses. The post processing of the stress plots is performed using the functionality provided by ANSYS. A sample result is depicted in Fig. 9.

Fig.9: Plot of calculated stresses 4. Conclusions The problem of handling a wide variety of different screw propeller geometries, user defined file formats and different CAD output files is overcome by using a simple ASCII file format for the propeller blade geometry. The Friendship Framework as back bone for the batch process and a series of different exchangeable calculation modules realize the steps from geometry to stress distribution, which can be evaluated by the approval engineer. The automated and faultless transformation of 3D boundary representation models could not reliably realised at this stage. Alternatively building up a 3D body from a number of unconnected surfaces does not solve the problem either. The created body is not a closed shape and can not be handed over to other tools without errors. However, it is of crucial interest to avoid manual geometry manipulation and transformation. For this reason a special solution has been developed and described. The concept of loosely coupling a modular chain of specialized tool proved to be very efficient. It is easily possible to develop or improve an individual component and integrate it to the tool by adding a simple code line to the controlling Python script. The future development of the work-flow will take advantage of the modularity. One of the first procedures to enhance will be the mesh generation. Additional modules will be elaborated in order to

377

deal with design variants such as the type of blade foots used in controllable pitch propellers. Here the blades are bolted to the hub and the critical stress areas may occur at the bolt holes due to notch effects. Currently a further development of the parameter definitions used for the propeller geometry definition is under discussion. The aim will be to satisfy more users and to enlarge the community of makers, institutions etc. being able to operate their systems with this kind of propeller definition. References GL (2011), Rules for Classification and Construction, I Ship Technology, 1 Seagoing Ships, Chapter 2 Machinery, Section 13, Germanischer Lloyd, Hamburg SCHULZE, R. (1998), Beschreibung des Propeller file-Formates (PFF file) Version 2.0, SVA Potdam, Report 2017 (rev.) BRUNK, H. (1998), Propeller Berechnung in der Klassifikation, Jahrbuch der Schiffbautechnischen Gesellschaft 92, Springer HARRIES, S.; KTHER, B.L. (1997), Rechnererzeugte Propellergeometrien, Jahrbuch der Schiffbautechnischen Gesellschaft 91, Springer SCHRDERL, J.(2003), Netgen 4.3, http://www.hpfem.jku.at/netgen/

378

You might also like