Methods of Longitudinal Weight Distribution
Methods of Longitudinal Weight Distribution
Hampton Roads Section of the Society of Naval Architects and Marine Engineers
Abstract
Approximation methods for weight distribution of ships are surveyed. Grouping methods such as the Bucket and station method are also explored. Detail based methods are explained. Finally, an improved method of distribution based on details is proposed. Guidance for the requirements of a weight database for this method is given and an alternative summary method is suggested to overcome difficulties caused by failure to meet certain database requirements of the detail method. Extensive appendices provide necessary figures and equations for using these methods.
Contents
Abstract.................................................................................................................2 Introduction ...........................................................................................................3 The Weight Distribution Problem ..........................................................................3 Approximation Methods ........................................................................................4 Grouping Methods ................................................................................................5 Direct Distribution Methods...................................................................................6 General Philosophy of Distribution ....................................................................6 Mechanics of Distribution ..................................................................................6 An Improved Direct Distribution Method............................................................7 Validating the Distribution ..............................................................................8 Database Requirements for Direct Distribution Methods...................................8 Summary Methods................................................................................................9 Accuracy of Weight Distributions ..........................................................................9 Conclusion ..........................................................................................................10 References .........................................................................................................10 Appendix A: Survey of Approximate Methods of Weight Distribution..................12 Appendix B: Equations for the Direct Calculation of Weight Distributions...........19
Introduction
Determination of the longitudinal weight distribution is vital to the proper calculation of the longitudinal strength of a ship. The longitudinal weight distribution also affects speed loss in a seaway [1]. Weight distributions of all three principal axes can also be used to calculate the ships gyradii [2] which have a profound effect on the seakeeping performance of a vessel. Before the advent of computers, determination of a ships weight distribution was a rather laborious process [3]. Due to the amount of labor involved, approximation methods were developed over the years. With the advent of computers, methods of collecting all of the weights with centers between given locations became less labor intensive giving rise to grouping methods. For longitudinal strength calculations, various levels of detail are acceptable. However, the standard is a Twenty Station Weight Distribution which actually consists of 22 weight segments divided by 21 stations, (Stations 0 through 20).
Approximation Methods
Numerous approximation methods for distributing hull weight have been proposed in the past. Hull weight is traditionally defined as lightship minus the weight of the anchor, chain, anchor handling gear, steering gear and main propulsion machinery. Determination of the exact breakdown of hull weight should be made based on the relative density of the object in question. Items left out of hull weight should be independently distributed as rectangles or trapezoids and combined with the hull weight distribution to determine the total weight distribution for the ship. Most approximation methods are based on combinations of a midship rectangle with forward and after trapezoids. More sophisticated methods base a portion of the weight curve on the ships buoyancy curve. Approximation methods are presented in works by Smith [3], Comstock [4], and Hughes [5] as well as in Principles of Naval Architecture [6]. Appendix A: Survey of Approximate Methods of Weight Distribution contains details and equations for several of these approximations. These approximations are general and appropriate only for initial stage design due to their low fidelity. An example of such an approximation is depicted in Figure 1.
Marine Vehicle Weight Engineering [7] contains longitudinal weight distributions by type for various military and support vessels. This segregation of distributions by type allows for improved fidelity. Such base distributions can be grossly modified to quickly arrive at a weight distribution of new designs similar to the provided types in the concept exploration stage of design. However, the resultant distributions are still not accurate enough for final longitudinal strength calculations.
Grouping Methods
The original grouping method is the Bucket method. The Bucket method derives its name from the fact that the weight details are metaphorically placed in buckets based on the location of their longitudinal center of gravity. If a line items longitudinal center of gravity falls in the extents of a bucket, it is included in that bucket. This method is illustrated in Figure 2.
The flaw of this method is that weight details are lumped representations of distributed weights. One line item can represent 1 or 1000 feet longitudinally. Thus just because the center reported in the line falls in one bucket does not necessarily mean that all or even the majority of the weight reported in that line falls in that bucket. Generally the most offending line items in such a method are items such as paint, weld, and mill tolerance weights as they generally reflect the weight of these items across the entire ship. Weights such as these are generally corrected by hand to improve the distribution. However, most other weights that belong in multiple buckets remain uncorrected. Distributed systems such as piping, electrical and ventilation systems often have weight records that have extents that span multiple stations. Thus the accuracy of this method is limited. The Ship Design Weight Estimate program used by NAVSEA utilizes an extra field in the weight record allowing the weight calculator to indicate whether the weight represented by that line resides in only the station it is in, is distributed over a number of stations about the records center, or is spread over the entire ship [8]. This station method improves the quality of the weight distribution and reduces the amount of rework needed to yield a reasonable distribution. It should be noted that this improvement only masks the inherent flaw of this approach; it does not eliminate it.
Mechanics of Distribution
The fundamental representative shape of direct distribution methods is the trapezoid. Representing a weight record as a trapezoid requires knowing the weight, the longitudinal extents, and the longitudinal center of the weight being represented. The computer program ShipWeight created by BAS engineering uses such an approach [9]. The fundamental problem with trapezoidal representations is that they are limited to weight records where the center resides in the middle one third of the length. Attempts to represent weight records whose center falls outside of the middle one third using the equations for trapezoidal representations result in part of the weight distributions being negative. Such an inverted trapezoid is portrayed in Figure 3.
Figure 3: The result of attempting to use a trapezoid to represent a record whose center is outside the center 1/3.
Such a representation is clearly flawed as it subtracts weight from a location that the record should be adding weight to. This can be overcome by requiring that the user adjust the inputs so that the center falls in the middle one third of the extents. This is accomplished by dividing the offending records into acceptable pieces or combining them with other records to create acceptable records and can involve a large amount of user interaction with the tool. Another option is to simply adjust the extents outward until the center is in the middle one third. Unfortunately, this option often reduces the accuracy of the resultant distribution. 6
As with the trapezoidal direct distribution method, the weight per foot of each weight record is summed for each location to determine the weight per foot curve for the entire ship. This is done at a high level of fidelity, say at every half foot, and then the resultant curve is summed to determine the 20 station weight distribution.
Summary Methods
Sometimes the requirements for direct calculation of a weight distribution discussed above are not met. Generally this is because the extents of each record are not included in the database or there is extensive use of impact records. These problems preclude the direct calculation of the weight distribution record-by-record. However, they do not render the general approach behind the direct method unworkable. The problems caused by impact lines and the lack of extents can be overcome through the use of summaries. In particular, summaries at the most detailed level under a given work breakdown structure (WBS) should be used. In the Expanded Ship Work Breakdown Structure used by the US Navy this is the 5 digit level. The use of these summaries requires user interaction. Discontinuities in a given WBS group must be represented by using a different weight record for each piece of a WBS group. The decision about how many pieces to use to represent a given WBS group is a matter of judgment. It should be considered that most WBS groups should be represented by at least two entries as many WBS groups have a coffin shaped distribution. Neglecting the discontinuities of WBS groups has varying effects on the accuracy of the final distribution. Only experience with this method can show the user how much effect a given choice will make. At early stages of a ship design, higher levels of abstraction in this summary method may be used to limit the effort required to prepare a distribution. For example, most weight groups can be roughly approximated by two summaries with extents based on the farthest forward and aft detail centers. This approximation will cause the length of the weight group to be underestimated. Such abstraction would raise the uncertainty of the distribution, but trading accuracy for speed of calculation may be preferable or necessary due to lack of detailed information during concept exploration or feasibility studies.
1/1000 of the ships length between perpendiculars. This value is the result of some sensitivity studies performed by the author and could be conservative.
Conclusion
Approximate methods can be useful in concept exploration and still have much to recommend them for early stage feasibility analysis. However, the improved direct method and its summary simplifications presented in this paper have attained a functionality that allows for far greater accuracy with a minimal increase in effort even at very early stages of design. Thus it is recommended that such approaches be used universally. Grouping methods such as the Bucket and station methods of weight distribution have been superseded by direct and summary methods of distribution. It is recommended that weight databases for new ship designs conform to the requirements for direct weight distribution as this would allow for rapid preparation of weight distributions on a regular basis through the ship design.
References
1. Swaan, W. A. and H. Rijken, "Speed Loss at Sea as A Function of Longitudinal Weight Distribution". International Shipbuilding Progress, Vol 11. No 115. March 1964. 2. Hansch, David L. Weight Distribution Method of Determining Gyradii of Ships. Paper No. 3399. Presented at the Chesapeake Bay Regional Conference of the Society of Allied Weight Engineers. Nov 2, 2006. 3. Munro-Smith, R. Applied Naval Architecture. American Elsevier Publishing Company, Inc. New York 1967. 4. Comstock, John P. Introduction to Naval Architecture. SimmonsBoardman Publishing Corporation. New York, New York. 1944. 5. Hughes, Owen. Ship Structural Design: A Rationally-Based, ComputerAided Optimization Approach. Society of Naval Architects and Marine Engineers. Jersey City, New Jersey. 1988. 6. Principles of Naval Architecture Volume I. Edited by Rossell, Henry E. and Lawrence B. Chapman. The Society of Naval Architects and Marine Engineers. New York. New York. 1941. 7. Marine Vehicle Weight Engineering. Edited by Cimino, Dominick and David Tellet. Society of Allied Weight Engineers, Inc. USA. 2007.
10
8. Filiopoulos, Christos and Debasish Ray, Total Ship Weight Management Computer Program For Todays and Tomorrows Applications. Presented at the 58th Annual Conference, Society of Allied Weight Engineers, Inc. 1999. 9. ShipWeight Users Guide Version 7.0. BAS Engineering. November 2005.
11
This sort of representation is typically used to approximate the hull weight, the steel, woodwork, fittings and outfit except anchors and cables, hull engineering except windlass and steering gear, any spread-out items of deadweight, such as passengers and crew, and designers margin. Comstock goes on to note that, The diagram must be proportioned that not only will the area be correct but also the LCG. The cargo should be, distributed over the length of the cargo holds as trapezoids, and so on until the diagram includes all the weights in the loaded ship.
12
This approximation is appropriate for passenger and cargo vessels. WH is the weight of the hull in tons and L is the length of the ship in feet. The centroid of the diagram (LCG) is given is 0.0056L abaft midships. The centroid can be shifted by increasing the ordinate at one end of the ship and decreasing the other. The amount to add and subtract (x) is defined as:
x=
54 WH Shift _ of _ Centroid * * L L 7
13
The Table below gives the ordinates for the plot based on Prohaskas work. A method to move the LCG from midships is not provided.
14
This distribution is intended for vessels which dont have parallel middle body. The centroid of the distribution can be shifted by swinging the parabola. This method is better depicted in the following figure from PNA [6].
As PNA states, Through the centroid of the parabola draw a line parallel to the base and in length equal to twice the shift desired (forward or aft). Through the point thus determined draw a line to the base of the parabola at its mid-length. The intersection of this line with the horizontal drawn from the intersection of the midship ordinate with the original parabolic contour determines the location of on point on the corrected curve. Parallel lines drawn at other ordinates, as indicated in Fig 4, determine the new curve.
15
Approximate Hull Weight Curve based on Buoyancy Curve from Hughes [5]
16
20 Station Distributions by ship type From Marine Vehicle Weight Engineering [7]
17
18
19
20
RH = WT
LDist
TH
2
2 RH + TH 2
H = TH +RH TRAP Slope = If FA = 1: H RH MAX CG MINCG Else: H RH MINCG MAXCG TRAP Intercept = If FA = 1: H TRAP _ slope MAXCG Else: RH TRAP _ slope MAXCG Center Check = If FA = 1 2 TH LDist Else: TH LDist
2
3 Weight _ check
+ RH LDist
2 + MINCG
6 Weight _ check
+ RH LDist
2 + MINCG
21
Virtual CZ = If FA = 1:
ZH = If FA = 1 WZ Local _ CZ LZ
LZ
Else
LZ
4
2
WZ Local _ CZ LZ LZ 2 6
2 LZ
RHC = WZ
LZ
ZH
Weight _ Z =
LZ RCH + ZH 2
Weight _ A = S LA
LZ
22
Z slope = If FA = 1 ZH LZ Else ZH LZ Z intercept = If FA = 1 RHC Z _ slope ( MINCG + LA) Else RHC Z _ slope ( MINCG + LZ ) Break Point = If FA = 1 MINCG + LA Else MINCG + LZ A Slope = If FA = 1 S LA Else S LA A intercept If FA = 1 S A _ slope ( LFWD + LA) Else S A _ slope ( LAFT LA)
23
Calculation Tool: The equations presented in this appendix enable creation of a spreadsheet that takes weight, extent and center inputs and calculates a representative weight distribution for each group. If the center of the group falls in the middle third of the groups length, the representation is a trapezoid. If the center is outside the middle third, a compound consisting of a triangle and a trapezoid represents the weight distribution. This calculator incorporates five compound combinations to represent shapes where the center is between 66 and 91.25 % of the length of the group from either extent. (The five compound shapes, applicable range, trapezoidal shape weight and the relative length of the triangular part of the compound appear in a table in the Equations section of this appendix.) The weight and length parameters were chosen to provide the greatest coverage.
After the distribution is calculated, the equation of the line along the top of the shape is calculated. The inputs and calculations described above all take place on a sheet labeled Entry. Sheets labeled A, Z, TRAP calculate the weight per foot from the three equations of the lines: the Sort sheet selects the correct weight for each location. This Sort sheet sums the total weight per foot at each location and then transfers this data to a sheet that stores the weight distribution. The weight is then calculated by Simpsons Rule and the center is calculated directly by summing the moments in order to verify the weight distribution. Segmenting and integrating the ships weight in this manner is accurate; however, the use of Simpsons Rule introduces slight integration errors. The difference in total ship weight is generally on the order of less than a half of a percent; this can be improved by increasing the number of samples taken along the axis.
24