Abstractâ The Smart Ship Systems Design (S3D) prototype is a comprehensive engineering and design environment capable of performing concept ...
Approach to Develop Ship Design Evaluation Rule-Base R.R. Soman, M. Andrus, M. Bosworth, I. Leonard, and M. Steurer Center for Advanced Power Systems (CAPS) Florida State University (FSU) Tallahassee, Florida, USA • “What are the systemic benefits of advanced cooling technologies?”
Abstract— The Smart Ship Systems Design (S3D) prototype is a comprehensive engineering and design environment capable of performing concept development and comparison (weights, power demand, speed, range, hull-form etc.), and high level ship system tradeoff studies. This online collaborative design environment is expected to be applied at the early stages of a ship design problem. Currently, the S3D environment contains tools for the development and simulation of the electrical, piping, and mechanical ship systems and the arrangement of the system and is capable of static power flow simulation for all major disciplines. However, the tool does not have a robust capability to evaluate designs using well established engineering guidelines. The research described herein aims to address this gap and this paper presents proposed approaches and outcomes of preliminary studies.
It is imperative that S3D, with its envisioned advantages centering on a collaborative, concurrent, multi-disciplinary design approach, be able to produce legacy ship designs as well as novel, innovative ideas, both with a minimum acceptable degree of standardization and risk assessment. A vital aspect of developing successful designs is the ability of the environment to evaluate them. The implications of the term “evaluate” are two-fold: 1) Adhere to standard practices – It is vital that the designer seek novel designs within the S3D environment. But, it is also equally important that established industrial/community norms and standards are followed. 2) Failure analysis – This is a slightly more complex capability as it involves expert understanding of faults/failures/disturbances with adequate mathematics to estimate useful life or devices and components. Such a capability would if used to test a design, would enable the user to de-risk the system.
Keywords—S3D; electric ship design; rule-base; NLP; FMEA
I. INTRODUCTION TO SMART SHIP SYSTEMS DESIGN (S3D) The concept design phase is arguably the most crucial phase of a product’s design, because errors or poor choices at this juncture can adversely impact any subsequent design efforts [1]. The US Navy recognizes the importance of the early stage design phases and the need for tools to facilitate them. In response to this need Electric Ship Research and Development Consortium (ESRDC) has aimed at developing, demonstrating, and testing the structure and feasibility of a collaborative environment that converge ship system designs at an early stage [2]. This impetus is with the intention to enable collaborative multi-disciplinary concurrent engineering where engineers working within their discipline’s particular tool can view and react to changes made by designers/engineers in other disciplines. The combination of concurrent engineering which parallelizes the design process to allow reduction in development time [3], and the capability where multiple designers work in tandem to achieve a better product [4], offers the possibility for rapid generation and analysis of ship concepts and ultimately greatly improved designs that feed into the preliminary design stage. The S3D environment is anticipated to enable the Navy to address early stage design related queries such as
It is important to note that at this stage of the research, the pertinent goal is to explore ways of enhancing existing capabilities and functionalities of S3D, by incorporating design evaluation functionality into S3D. An ultimate goal for the future could well be to develop a comprehensive S3D environment with detailed analysis functionalities with accurate design evaluations for electric ships The importance of product model data at early stage ship design is emphasized within the Naval Sea Systems Command (NAVSEA) as being critical to enable timely, efficient and accurate response. The product model repository used by NAVSEA is called Leading Edge Architecture for Prototyping Systems or LEAPS [5]. Further, importance is also placed on an extensible-information metamodel to support modeling and simulation tools used by navy ship designers with the current focus on concept studies, analysis of alternatives and operational scenarios along with the ability to link to external analysis tools as shown in Fig. 1.
• “How much energy storage is needed for a given design?” This work was sponsored by the US Office of Naval Research under contract N000141410198
978-1-4799-1857-7/15/$31.00 ©2015 IEEE
193
and provide preferences and solutions on their own perspectives. • Comparative analyses – Intersection between sets are used to establish feasibility before an optimal solution is finalized. • Evolving complexity – Fidelity of analysis is increased as the design progresses. Fig. 2 shows the SBD design stages involving a concurrent approach from an initial “separate solution” stage to a final “optimal solution” stage. Such an SBD based approach was used for the first time in a ship design and acquisition program for the US Navy under the Ship to Shore Connector Program in 2007. McKenney and Singer [6] elaborate in detail about using such an SBD approach for ship design with perceived issues and probable solutions.
Fig. 1 LEAPS product model repository linked with external tools to support virtual prototyping of complex naval systems at NAVSEA [5].
This paper discusses the general and widely used ship design methodologies relevant to the medium voltage DC (MVDC) system and attempts to merge it with development efforts pertaining to S3D. The work done regarding this task revolves around applying systematic approaches that would enable: • Extracting ship design guidelines – Initially for the early, conceptual design phase at the system level advanced, and eventually at a more device/component level stage. • Facilitate risk assessment – Using well known engineering and statistical equations to predict when a component or device and hence a system might fail/malfunction. Another vital aspect is the possibility of adapting established risk assessment techniques such as failure mode and effects analysis (FMEA) and fault tree analysis (FTA) in order to integrate it as a design evaluation functionality into S3D. • Provide decision support – Tap into information provided through design guidelines and failure assessment to present the designer with recommendations that would improve the design.
Fig. 2 Concurrent systems engineering process [6]
A. Concurrent design application to ships- Set based design The essence of S3D as a design environment is the ability for designers across different analyses domains to simultaneously access data to iteratively modify and evolve the overall design. This parallels the well-known engineering realm of concurrent engineering. Set-based design (SBD) falls into such a category, with specific application to large team-based complex design spaces [6]. The similarities between SBD and the envisioned S3D functionalities could be summarized as follows:
B. Generic Electric Ship Design Process Fig. 3 displays the classical system engineering process [7]. It illustrates a highly iterative design process in which a set of “direct”, or operational requirements for a desired system design, i.e. the “System Need”, spawn additional “derived” requirements as they undergo functional allocation analysis, and design synthesis operations. Taken together, these direct- and derived requirements form the engineering rules-base for the intended system design.
• Numerous choices – Large number of design alternatives are considered by exploring the design space. • Information exchange – Separate teams of specialists (designers) are able to evaluate outputs
194
The general design methodology is shown to be the product of a generalized process for designing electric warships proposed by Doerry [8] and consisting of the following steps:
anticipated outcomes) seeks to allocate specific design requirements (both direct- and derived) to mission components that fulfill a specific function in the overall process of providing electric power to the ship’s loads. Analyzing a desired ship power system design on the basis of the functions it is intended\required to perform can also help define the rules-base for the system. The process proposed involves performing a top-down, “functional decomposition” of the system. Doerry [7] performs a functional decomposition of a MVDC integrated power system (IPS) in which the principal function of the power system is to, “Safely generate, transport, and deliver electrical power of the proper quality and continuity needed by the served loads.” The list of common functions of the IPS includes, but is not limited to, the following [7]:
1) Analyze requirements, 2) Allocate requirements to mission systems, 3) Develop initial Concept of Operations (CONOPS), 4) Assign mission systems to ship zones, 5) Develop derived requirements for ship systems, 6) Develop distributed system architectures, 7) Calculate distributed system component ratings, 8) Synthesize the ship, 9) Evaluate total ship mission effectiveness, and 10) Iterate until total ship mission effectiveness requirements are met.
1) 2) 3) 4) 5) 6) 7) 8)
Power Management – Normal Conditions, Power Management – Quality of Service, Power Management – Survivability, System Stability, Fault Response, Power Quality, Maintenance Support, and System Grounding.
The following are some of the system requirements/design rules inherent to the definition of these functions: 1) Under normal operating conditions, the IPS shall be capable of being configured such that: a) All loads receive sufficient electrical power, b) Sufficient rolling reserve be provided to supply load steps due to pulse loads, large motors starting and large radars changing modes of operation (e.g. cruise to battle), c) Balance between the average power generated and consumed/dissipated is maintained. (e.g. total power generated = total power consumed/lost as heat), d) Dedicated energy storage is installed in order to level load spikes, e) Dedicated energy disposal is installed in order to level load spikes, f) Expected system dynamics do not cause any of the energy storage mechanisms to either “overfill” or “run dry”, g) The use of energy disposal is minimized, h) Generators operating in parallel shall share load power without requiring dedicated communication lines, i) System stability is maintained during system disturbances,
Fig. 3 Classical systems engineering process [7]
The principal focus of the study performed under this project was to provide the ship designer with information and tools for completing steps 1) and 5) for the design of the baseline MVDC ship electrical system. The section that follows describes the process of defining the “rules-base” for a hypothetical ship system. This report details the preliminary work for extracting ship design guidelines and rules from well-known resources. Furthermore, the guidelines and rules extracted through this task will be integrated into the S3D environment to enable applying well established engineering principles to facilitate more automated design evaluations. However, the integration of design guidelines into S3D is an ongoing activity and out of the scope of this paper. This document reports on details, benefits and outcomes of the approach used to extract design rules. C. Direct and derived requirements for MVDC shipboard power systems The functional analysis allocation operation in the classical system engineering process (Fig. 4 Individual aspects feeding the overall methodology leading to
195
j) During the initial five minutes following an imbalance between electrical power generated and power consumed and dissipated, the IPS shall be capable of ensuring that the Quality of Service (QOS) standards [9] for all loads are maintained. 2) Under conditions where the power system cannot serve all loads, due to either battle damage or equipment failure, the IPS shall
requirements for inclusion in Step 5) of the ship design process. A detailed description of the actual NLP-based tools used in the study is provided in Section II. II. NLP AND FMEA AIDED APPROACH The proposed approach includes investigating IEEE-Std. and MIL-handbooks, for which natural language processing (NLP) based ideas were used in the initial stages. NLP [10] is an exhaustively researched wing of artificial intelligence mainly dealing with creating “intelligent” human-machine interfaces focusing on increasing the levels of automation related to computational tasks. In this case, NLP principles were used to heavily reduce the man-hours that would have gone into investigating technical resources. Following subsections show an example of NLP based data-mining outputs for IEEE Std.C57.91-2011 and the process ‘identified pages’ as ’worth investigating‘, along with the information found on such highlighted pages. This particular IEEE standard is a guide applicable to loading of transformers and is chosen as a sample for demonstrating the approach proposed in this paper Fig. 4 illustrates this proposed approach that could potentially lead to enable identification of intermediate and advanced MVDC shipboard power system design rules and recommendations related to the risk assessment of potential designs. Subsequently, these results are perceived to be the input to S3D, as a framework of design rules and recommendations with risk assessment functionality. The approach begins with applying the NLP data-mining techniques, described herein, to the body of IEEE and Military standards and handbooks dealing with relevant information regarding MVDC shipboard power system design. The objective is to extract as much risk-related information as possible that could be deemed necessary from a design point of view (i.e. risk-related, system design requirements and rules).
a) Exhibit a proper survivability response, e.g. appropriate loads are shed in the order of their mission priority, b) Determine the health of loads and power system equipment, c) Restoring power to shed loads if sufficient capacity and connectivity is present, and the load is safe to reenergize, d) Isolating unsafe loads, Be capable of optimal reconfiguration following a disturbance. 3) System Stability of the IPS shall ensure that a) Maintaining system stability in the presence of negative incremental resistance on the DC bus, b) Generator speed is not directly observable on the DC bus, c) The kinetic energy of the prime movers is neither too low to cause the device to stall/shutdown, or the voltage to collapse, nor too high to cause an overspeed induced shutdown or failure. D. Requirements and rules search using natural language processing Step 1) of the ship design process in section I.B involves analyzing pertinent Navy/IEEE policy, practices, customs, statutory, standards, and engineering guidelines documents to identifying operational requirements (i.e. design rules and experience-based guidelines), as well as the derived/synthesized requirements from design step 5). In order to make such a tedious task easier, a natural language processing (NLP) [10] tool was developed and used during the literature search phase of the work. NLP algorithms are capable of performing exhaustive searches of IEEE/MIL standards and handbooks for rules-related terminology. IEEE-Std.1709 [11] identifies specific standards directly relevant to MVDC shipboard power systems, each of which are investigated by a team of experts in order to extract pertinent information that could be termed necessary from a design point of view. The relevant resource referred to in this paper is IEEE Std. C57.91-2011 [12], which deals with performance and prevalent practices for transformers. An example is described in Section III that shows the use NLP to identify component-level design rules that have the potential of impacting the design of the ship power plant at the system-level. The design equations for both examples are analyzed and shown to generate new (i.e. derived)
Fig. 4 Individual aspects feeding the overall methodology leading to anticipated outcomes
An effective FMEA presents an examination of a system’s strengths and weaknesses [13] - [16]. This assessment could be done in one of two ways; via functional
196
FMEA (F-FMEA) or hardware FMEA (H-FMEA) [17]. As FMEAs are best initiated during the conceptual design phase (long before specific hardware information is available), the functional FMEA approach is generally the most practical and feasible method by which to begin. This is especially true for large, complex systems that are more easily understood by function than by minute details of their operation. When systems are highly complex, the analysis for F-FMEAs generally begins at the highest system level and follows a top-down approach. H-FMEAs typically begin at the lowest piece-part level and use a bottom-up approach. Prior work relating to F-FMEA and H-FMEA relevant to shipboard power systems is explained in [18], and [19].
Network topology information (advanced stage application in S3D) This knowledge is useful to the monitoring system, which could be fed to supervisory control architecture for more informed decision support. This application too, is advanced for the current stage of this work. FTA is another risk assessment technique in which the probabilities of a fault event are assigned to compute the overall probability of failure. Though slightly advanced than an FMEA, FTA could be adapted to be used as part of an early stage ship design process owing to the availability of failure probability data in MIL-handbooks. This paper explains the overall approach initiated through NLP and resulting outcomes with inclusive FMEA aspects. The next subsection attempts to provide a sufficiently detailed overview of results and their utilization with some practical observations.
A. Potential uses of general FMEA outcomes FMEA in general follows the methodology of breaking down a system into smaller functional parts. In other words, an overall network is broken down into devices and devices, in turn, into components. Then, a piece by piece failure analysis is initiated, keeping in mind the local and global effects of the considered failures along with causes. This piece-wise approach helps build up fault related understanding at various complexity levels of a system. The following points summarize the outcomes of a detailed FMEA and its relevance to this research task:
B. Snapshot of results using NLP on [12] Generating results As mentioned in section I.D, IEEE Std.C57.91-2011 is the major design related reference used to form guidelines for a transformer evaluation. This subsection succinctly explains results for IEEE Std.C57.91-2011. This research uses code developed in statistical software package R and its NLP functionalities, specifically from information contained in IEEE Std.C57.91-2011. TABLE I shows the top six results for most keywords on a page for IEEE Std.C57.91-2011. A different sorting scheme produces results per keyword as shown in TABLE II for the same text. Both these views provide direct information that point to important pages in the text. TABLE III shows the [20] measure and sorts the pages as per decreasing value of this metric. In other words, it indicates pages as per importance based on the keywords it contains. As an example referring to TABLE III, page #18 is likely to contain more relevant words as compared to page #33 since its value is higher.
Power quality (intermediate stage application in S3D) Single out system areas which could have power quality issues due to multiple power conversions. This knowledge is useful to judge which zones may be most critical from the power quality point of view. It also helps evaluate the S3D user’s choice of power conversion equipment which is vital to the benefits of the integrated zonal distribution approach. Zone priority (intermediate stage application in S3D) This is done based on the importance of constituent loads within a zone. The knowledge about mission specific vital loads may prove useful during operations like load shedding. This potentially ties in with the naval systems engineering design paradigm to carry out system-wide tasks under specific conditions and maintaining standards. In this sense, the rank of the zone would determine the conditions and standards that define it, in turn dependent on its constituent loads.
TABLE I.
TOP SIX PAGES AS PER KEYWORD COUNTS
keyword count
294
249
247
242
241
227
pg#
112
115
69
38
17
42
The results across TABLE I to TABLE III help the researcher focus the search and quicken it in at least 2 ways:
Intra-zonal risk assessment (advanced stage application in S3D) More detailed analysis into sub-sections within a zone gives insight into breaking the system into smaller parts for a more exhaustive failure study. This knowledge is useful when trying to apply well known traditional diagnostic methods with modifications if needed which may in turn aid in the diagnostic technique selection process. This is an advanced stage application perhaps at a more sophisticated level of S3D development.
Frequency metric (simple) This approach is based on how many keywords appear on a page. This is a relatively simple measure but helps relegate pages with lesser number of keywords while the ones with the most keywords are identified. Weighted metric (complex) With the output of keywords, this metric weighs them according the measure, thus providing a second
197
perspective that’s relatively more complicated. Both in tandem are a reliable methodology for the researcher to focus their search for the most relevant pages of a resource thus greatly reducing time spent and increasing efficiency.
TABLE II and TABLE III. TABLE I indicates page #17 to be important. TABLE II corroborates this output by identifying the keyword ‘insulation’ appearing on page #17. TABLE III.
Utilizing and applying results The major advantage of using the approach based off NLP is the reduction in man-hours spent to read resources. The results readily produce a list of pages from a text resource which are likely to contain useful information pertaining to the following broad categories: Limits and tolerances General minimum and maximum values of engineering applications such as: • Temperature • Pressure • Dimensions • Electrical quantities
Frequency
, % , + , ± , ac , dc , acdc , dcdc , dcac , acac , acdc , dc-dc , dc-ac , ac-ac
Fig. 5 Design rules to evaluate transformer selection for chosen loading profile
[10] E. Cambria, and B. White, “Jumping NLP Curves: A Review of Natural Language Processing Research [Review Article],” IEEE Computational Intelligence Magazine, vol.9, no.2, pp.48-57, May 2014. [11] “IEEE Recommended Practice for 1 kV to 35 kV MediumVoltage DC Power Systems on Ships,” IEEE Std. 1709-2010, vol., no., pp.1, 54, Nov. 2 2010. [12] “IEEE Guide for Loading Mineral-Oil-Immersed Transformers and Step-Voltage Regulators,” IEEE Std. C57.91-2011 (Revision of IEEE Std. C57.91-1995), vol., no., pp.1, 123, March 7 2012. [13] J.M. Legg, “Computerized Approach for Matrix-Form FMEA,” IEEE Trans. on Reliability, vol.R-27, no.4, pp.254-257, Oct. 1978. [14] S.A. Herrin, “Maintainability Applications Using the Matrix FMEA Technique,” IEEE Trans. on Reliability, vol.R-30, no.3, pp.212-217, Aug. 1981. [15] H.B. Dussault, “Automated FMEA Status and Future,” in Annual Proc. of Reliability and Maintainability Symposium, 1984, vol., no., pp.1-5, 1984. [16] P.L. Goddard, “Software FMEA techniques,” in Annual Proc. of Reliability and Maintainability Symposium, 2000, vol., no., pp.118-123, 2000. [17] R.J. Mikulak, R. McDermott, and M. Beauregard, The Basics of FMEA, 2nd ed., CRC Press. [18] RR. Soman, E.M. Davidson, and S.D.J. McArthur, “Using functional failure mode and effects analysis to design the monitoring and diagnostics architecture for the zonal MVDC shipboard power system,” in Proc. of the IEEE Electric Ship Technologies Symposium, 2009\ vol., no., pp.123-128, 20-22 April 2009. [19] RR. Soman, E.M. Davidson, S.D.J. McArthur et al, “Modelbased methodology using modified sneak circuit analysis for power electronic converter fault diagnosis,” in IET Trans. on Power Electronics, vol.5, no.6, pp.813-826, July 2012. [20] Fei Liu, Feifan Liu, and Yang Liu, “A Supervised Framework for Keyword Extraction From Meeting Transcripts,” IEEE Trans. on Audio, Speech, and Language Processing, vol.19, no.3, pp.538,548, March 2011.
ACKNOWLEDGEMENT The authors would like to thank Dr. Roger A. Dougal and his research team at the University of South Carolina (USC) for their support and contributions in this project.
REFERENCES [1]
[2]
[3] [4] [5] [6]
[7]
[8] [9]
Q. Wang, M. Rao, and J. Zhou, “Intelligent systems for conceptual design of mechanical products,” in. Handbook of expert systems applications in manufacturing: structures and rules, New York: Chapman & Hall, 1994, A. Mital, and S. Anand, editors. J. Chalfant, B.Langland, and S. Abdelwahed et al, “A Collaborative Early-Stage Ship Design Environment,” (ESRDC library, 2012). Available on: http://www.esrdc.mit.edu/library/ESRDC_library/Chalfant_120 5_1.pdf. B. Prasad, Concurrent engineering fundamentals: Integrated product and process organization. Prentice Hall 1996. W. Shen, Q. Hao, and W. Li, “Computer supported collaborative design: Retrospective and perspective,” Computers in Industry 59 (2008) 855–862. Available on :http://www.navsea.navy.mil/nswc/carderock/docs/leapsOvervie w.pdf. T. McKenney, and D. Singer, “Set-Based Design: A concurrent engineering approach with particular application to complex marine products,” Marine Technology, July 2014, www.sname.org. N.H. Doerry, and J.V. Amy, “Functional Decomposition of a Medium Voltage DC Integrated Power System,” in Proc. of the ASNE Shipbuilding in Support of the Global War on Terrorism Conference, Biloxi, MS, April 14-17, 2008. N.H. Doerry, and H. Fireman, “Designing All Electric Ships,” in Proc. of the 9th International Marine Design Conference, 1619 May, 2006, Ann Arbor, MI, USA. N.H. Doerry, and J.V. Amy, “Implementing Quality of Service in shipboard power system design,” in Proc. of the IEEE Electric Ship Technologies Symposium, 2011, vol. no., pp.1,8, 10-13 April 2011.
200