A Scalable Search Algorithm for Unstructured Peer-to-Peer Networks
A Scalable Search Algorithm for Unstructured Peer-to-Peer Networks
Amit Gud Masaaki Mizuno Daniel Andresen
Department of Computing and Information Sciences
Kansas State University
234Nichols Hall,Manhattan KS,66506,USA
Abstract
Scalability in a peer-to-peer network is a challeng-ing problem.Unstructured peer-to-peer networks inher-ently lack scalability,and structured networks are inef-?cient for a high churn rate.In this paper,we present a scalable search algorithm for a decentralized un-structured peer-to-peer network using a method to dy-namically determine the number of nodes to forward a query to at once.The decision is based on the de-gree to which each neighbor has contributed to previous successful searches.The algorithm automatically cre-ates a spanning graph of the high traf?c links.Once a sta-ble spanning graph is created,a query tends to travel along the edges of the spanning graph.This way,the num-ber of hops required for a search is roughly bound by the di-ameter of the spanning graph.The simulation shows that our algorithm demonstrates signi?cantly better perfor-mance in terms of the number of messages generated and hops required for a search over other popular algo-rithms.
Keywords:P2P,resource discovery algorithms,dis-tributed systems performance
1.Introduction
Over the past decade peer-to-peer(P2P)network has shown rapid and signi?cant growth along with the growing Internet traf?c.P2P networks currently take a major share of the Internet bandwidth.The popularity of the P2P net-work has been triggered by the need of?le-sharing over the network,which made the study of performance,and scala-bility of such networks an interesting area of research.
P2P networks are classi?ed based on the topology that they create and their degree of centralization.A central-ized P2P network is the one which has a dedicated central server for maintaining the information of the?les shared along with their locations.This architecture is vulnerable to single point of failure.Napster is an example of such a net-work[9].
Decentralized P2P networks overcome this drawback by not having single central server.In a decentralized P2P network,no node has any special capabilities and all the nodes behave in the same manner.Decentralized networks are further classi?ed as structured and unstructured.In structured networks,the object1placement is determinis-tic which helps in satisfying a query2easily.Such struc-tured networks are con?ned only to the research literature and are practically non-existing in the real networks[3],[6] and[7]are examples of structured network.
In decentralized unstructured network,the precise loca-tion of an object is not known.Since this topology is more in use today there is a need to improve on the search mech-anism of this architecture.Gnutella is an example of such a network.Further details of the classi?cation of P2P can be found in[3].
A decentralized unstructured P2P overlay3network is built on top of the existing network topology.This is similar to building a connected graph,wherein every node knows only about a limited set of other nodes on the graph.In such networks,the precise object location is not known.A search technique has to be deployed in order to inquire at each node present in the network about the presence of the object. Early search mechanisms primarily used?ooding[1]or k-random walk[2]algorithms.In the?ooding approach, each node propagates the query to all its neighbors.On a re-ceipt of a query,the node searches in its local repository.If the object is found,it informs the query originator and fur-ther search in that path terminate.If not,the node further forwards the query to all its neighbors.To avoid duplicate query servicing,each node maintains a list of all the queries that it has serviced.As the number of nodes scanned per hop
1The terms object and?le are used interchangeably throughout the pa-per.
2Throughout this paper,query signi?es an object/?le search in the P2P network and the terms query and search are used interchangeably. 3An overlay network is built on top of the existing real network by log-ically connecting the involved nodes.
is high,the search time(i.e.,the number of hops/query)re-quired is reduced.
The?ooding method generates a large amount of net-work traf?c.To overcome this problem,random walk al-gorithms are often used.In the random walk search algo-rithm,the number of nodes to which a query is forwarded by each node N i is limited to a number k[2],where value of k is between1and NB i,where NB i is the number of known neighbors of node N i.These k nodes are selected randomly.Though highly random and non-deterministic, this approach is demonstrated to work more ef?ciently than ?ooding[4],[8].
This technique uses feedback from the previous searches which helped probabilistically guide the future ones.The grade of a node is determined by this feedback information, which indicates the possible success rate.
This paper presents an algorithm for improving the ob-ject search mechanism in decentralized unstructured P2P network by using heuristics based on the con?guration,per-formance and historical contribution of a node with respect to successful object searches.
Our search method,Activity Based Search(ABS),im-proves the previous probabilistic search techniques by mak-ing the k value adaptive4.Under ABS,each node keeps a list of known neighbor nodes sorted by the degree to which each neighbor has contributed to the previous successful search(such degree values are called“activity levels”).Ev-ery query is sent to the top k d nodes.A distinctive feature of our algorithm is that,unlike[11],a node does not consider the exact semantic preference when choosing the nodes to forward a query to.A node simply considers the previous success ratio,regardless of the objects being searched,to determine a set of nodes to forward a query to.Thus,its scalability is better than that of APS in terms of the stor-age size.
Simulations show that our algorithm automatically cre-ates a spanning graph of the nodes fairly quickly(see Fig-ure1),where the nodes in the graph represents the nodes in the network and the edges in the graph represents a sub-set of the links in the network.The communication traf?c on the edges in the spanning graph is substantially higher than that on links not in the spanning graph.Ef?cient creation of such a spanning graph is realized by the dynamically adapt-ing value of k d and selecting nodes with the highest activity levels to forward a query to.
The performance simulation demonstrates that our algo-rithm demonstrates signi?cantly better performance than k-random walk and1-random walk.Our messages-per-query is much less than that of k-random walk,yet our number
4To indicate that our k value dynamically changes,we use notation k d to denote our k value.of hops to reach the target nodes is much less than that of 1-random walk and even slightly less than k-random walk. The organization of the paper is as follows:related pre-vious work is brie?y described in Section2,our algorithm is presented in detail in Section3,and the experimental re-sults are presented in Section4.We conclude in Section5.
2.Previous Work
Random walk search algorithms have demonstrated to perform well in unstructured decentralized P2P networks [4].However there has been attempts to further improve the basic random walk algorithms.These works try to improve the random walk algorithms either by using feedback from the previous search results[5],[10],[8]or by constructing different overlays depicting node-level semantics,e.g.tem-poral and geographical localities[11],[12],[13],[14]. The feedback mechanism is used in building a pro?le of the known neighboring nodes.When a query has to be forwarded further,the node uses these pro?les to rank the neighboring nodes.The feedback information is maintained in the form of grade for each node.This grade is then in-creased or decreased depending upon the result of the ob-ject search passing through the node.Thus both learning and unlearning characteristics are exhibited by the nodes. The query is forwarded to top-performing k nodes,i.e.the nodes having top k grades.
Dimitrios et.al.[5]discusses such a method.In this,only k walks are generated,i.e.,the query originator forwards the query to k nodes in the?rst level of search.If the object is not found in the local repository,the node further forwards the query to only one top performing node.So from the sec-ond level onwards the query is forwarded only to one node. In-effect this limits the search within the k paths.Further-more,the value of k is?xed at each node and is not adaptive to dynamically changing network sizes.The k value of our ABS algorithm,denoted the k d value,is dynamically ad-justed to improve search performance.
[11],[13]and[14]present a technique to make the net-work scalable by exploiting the common interests of the peer nodes.Their protocol has demonstrated to improve the search results.However,they emphasize creation and main-tenance of the semantic overlays,rather than enhancing the scalability of the search.Our approach dynamically adjusts the degree of?ooding to make the search more scalable. Chawathe et.al.[15]suggests a topology-aware P2P sys-tem,which dynamically adapts to the changing network topology.This involves balancing the degrees of a node pro-portional to its capacity as de?ned by its access bandwidth, processing power and disk speed.Search mechanism,how-ever,does not attempt to scale with the network size and the number of nodes involved in an object search is static.node
(a)75
queries.(b)250
queries.(c)500queries.
Figure1.Connectivity graph for10nodes.Note similarity of overall patterns between graphs,indi-cating each node has formed a short list of“preferred partner,”essentially forming an implicit span-ning graph.Edges shown have the heaviest(top25%)activity levels.
does not weigh much,as much as the perceived track record of a node.
3.The Activity Based Walker algorithm
When a node,say N I,tries to search a?le F T in the set of nodes U,N I initiates a query message.The query mes-sage is passed among nodes.When a node which owns F T, say node N T,receives the query,it informs N I by initiat-ing a“success”message.In the above scenario,N I,F T,N T are called the initiator of the query,the target?le,and a tar-get node,respectively.
Each node N i maintains a set of?les(names or refer-ences)in data structure F ILES i.When a node N i receives a query from a node N j,N j is called the sender of N i. Upon a receipt of the query,N i checks whether it has the target?le by referring to F ILES i or whether it knows a node which owns the target?le(the later case is discussed below).If so,it initiates a“success”message;otherwise,it forwards the query to a set of nodes that N i knows.Our pro-tocol has salient features in how to determine a set of nodes to forward a query to.
To determine a set of nodes to send a query to,each node N i maintains,in data structure NBRS i,a list of neighbors, that is,the nodes that N i knows.NBRS i is a?nite list of tuples[node N k(its address such as IP),its activity level (denoted AL k,discussed later),and F ILES k],ordered by the activity levels.Since NBRS i is?nite,when a new tu-ple is added,a existing tuple with the lowest activity level may have to be dropped from the list.
The activity level AL j of node N j in NBRS i is a value calculated based on the success ratio in the previous searches;that is,the ratio at which N j was on the paths to the target nodes in the previous searches,regardless of the target?les.AL j in NBRS i is increased by some pre-de?ned value,denoted AL INC1,if the query sent to N j leads to a target node.In addition,if N k is a target node of the search,AL k of N k’s tuple in NBRS i is increased by some prede?ned value AL INC2;or if N k’s tuple is not in NBRS i,a new tuple[N k,AL INIT,F ILES k]is added to NBRS i,where AL INIT is a prede?ned initial activ-ity level.All activity levels of NBRS i are between0and 1,normalized with respect to the highest activity level in NBRS i.
When initiating a new query or upon a receipt of a query, the number of nodes to which node N i forwards a query at once dynamically changes by the following three factors.
(1)a prede?ned value k-threshold,which is the lower bound on the activity levels of nodes to which the query is for-warded,(2)a prede?ned value k-min,which is the lower bound on the number of nodes to forward the query to,and (3)the number of nodes in NBRS i that the query has not yet visited.Since each query contains a list of nodes that it has already visited,it is easily known whether the query has already visited a given node.However,this does not mean that a node is not visited more than once in one search.
A query branches out to k d queries at each node,result-ing many different search paths.However,a list of visited nodes is only maintained in each message;therefore,the list in a query message may not contain a node which is visited by a query message on a different search path.
When initiating a new query or upon a receipt of a query, node N i sends the query concurrently to all nodes N j such that(1)AL j≥k-threshold or(2)N j’s tuple is in the top k-min among the nodes in NBRS i that the query has not yet visited.We refer to this the search-node-determination al-gorithm,and the number of query messages to send deter-mined this way is called the k d value.
3.1.Success and failure of a search
The F ILES j?eld of each node N j in NBRS i works as “one-hop replica”,that is used to accelerate the search pro-cess.When a node N i initiates a new query or receives a
query for a target?le F T,it tries to locate F T not only its own F ILES i,but also F ILES j in all tuples in NBRS i, before forwarding the query to its neighbor nodes.
A node N i claims a search for a target?le F T to be“suc-cess”if it locates F T in F ILE i or in its one-hop replica.If N i is not the initiator of the query,N i initiates a“success”message and sends it to its sender N j.The“success”mes-sage contains N i(the target node)and F ILES i.N j then forwards the“success”message to its sender.This process continues until the“success”message reaches the initiator of the query.All nodes on the path,say N j,update N i’s tu-ple in NBRS j in the manner discussed above.
A node N i claims a search for a target?le F T to be“fail”if(1)it cannot locate F T in F ILE i or its one-hop replica, and(2)all nodes in NBRS i to which N i could forward the query have claimed the search to be“fail.”Then,N i initi-ates a“fail”message and sends it back to the sender N j. If N i has received“fail”from all nodes that it had sent the query to,it determines a new set of nodes in NBRS i by ap-plying the above search-node-determination algorithm and sends them the query.This process repeats until either N i receives“success”or there is no more node in NBRS i to try(in such a case,it claims the search“fail”).
If a query visits a node which has already been visited by a query on a different search path,the node immediately returns“fail”to its sender.
In order to ef?ciently terminates a search process,we ap-ply a technique called checkback[2]A query message con-tains an integer?eld called LiveCount.When the query vis-its a new node N i,N i decrements LiveCount by one.When the count becomes zero,N i communicates with the ini-tiator by sending a“checkback”message.Upon a receipt of a“checkback”message,the initiator decides whether it should continue or terminate the search and informs N i accordingly.If the search is terminated,N i just stops the search process;otherwise,N i resets the LiveCount to a pre-de?ned value,denoted LC INIT,and continues the search.
3.2.Node join
When a new node N i joins,it must know at least one node in the system.Let N j be such a node.N i?rst sends a“join”message with its F ILES i to N j.N j then replies N i with a“join ok”message with NBRS j and F ILES j. N j adds a tuple[N i,AL DEF,F ILES i]in NBRS j,where AL DEF is a prede?ned default initial activity level.N i adds a tuple[N j,AL DEF,F ILES j]in NBRS j and uses it as its own neighbor list NBRS i.
3.3.Protocol description
The query message consists of the following Fields:(1) N I(initiator),(2)F T(target?le),(3)PATH(a list of all nodes that the query has visited),and(4)LiveCount.
The initiator N I executes the following algorithm when searching target?le F T:
var
search:list of tuples;
outstanding,found:set of nodes;
if(F T∈F ILES I)return{N I};
if(F T∈F ILES j such that tuple
[N j,?,F ILES j]∈NBRS I)return{N j};
search=NBRS I;found=?
while search=?do{
outstanding=set of nodes selected by
search-node-determination algorithm;
search-=outstanding;
send query{N I,F T,P AT H={N I},LiveCount=LC INIT} to each node in outstanding;
while outstanding=?do{
receive a reply from a node N j;
outstanding-={N j};
if the reply from N j is“success”{
add AL INC1to AL j in NBRS I;
if target node N T is in NBRS I
add AL INC2to AL T in NBRS I
else
add[N T,AL INIT,F ILES T]to NBRS I
Normalize the activity levels in NBRS I
found∪={N T}
}
}
}
return found;
When a node N i receives a query{N I,F T,PATH,LiveCount} from N S,it executes the following:
var
search:list of tuples;
outstanding,found:set of nodes;
isSuccess:boolean;
if(--LiveCount==0){
send“checkback”to N T;wait for a reply from N T;
if(N T requests the search to be continued)
LiveCount=LC INIT;
else terminate the search
}
if(F T∈F ILES i)send N S“success”with{N i,F ILES i};
if(F T∈F ILES j such that[N j,*,F ILES j]∈NBRS i)
send N S“success”with{N j,F ILES j};
search=NBRS i?P AT H;isSuccess=false
while search=?do{
outstanding=set of nodes selected by
search-node-determination algorithm;
search-=outstanding;
send query{N I,F T,P AT H+={N I},LiveCount}
to each node in outstanding;
while outstanding=?do{
receive a reply from a node N j;
outstanding-={N j};
if the reply from N j is“success”{
add AL INC1to AL j in NBRS i;
if target node N T is in NBRS i
add AL INC2to AL T in NBRS i
else
add tuple[N T,AL INIT,F ILES T]in NBRS i
Normalize the activity levels in NBRS i
isSuccess=true;
forward“success”to N S;
}
}
}
if(?isSuccess)send“fail”to N S;
ABS succeeds by actively grading the nodes in the NBRS lists.This grading,called the activity level,is based on previous performance of the node and helps de-termine the level of?ooding;that is,the number of nodes to which a query is forwarded depends on the activity level of a node.The parameter k-threshold is used to con-trol trade-of between the amount of?ooding vs.number of hops required to complete a query.The higher the value of k-threshold,lower is the k d value,and thus less num-ber of messages.This serves to minimize free-riders, and helps create a spanning graph of search paths ef?-ciently(Figure1),and results in ef?cient communication in?nding the target nodes.
4.Experimental results
We compare the performance of our algorithm with two types of random-walker algorithms:k-RW ALK and1-RW ALK,using a well-known discrete event simulator for networks,Network Simulator2(NS2)[16].The difference between k-RW ALK and1-RW ALK is that in k-RW ALK,the initiator as well as all intermediate nodes send a query con-currently to k randomly selected nodes in their NBRS lists, whereas in1-RW ALK,only the initiator sends a query to k randomly selected nodes and other nodes forward the query only to one randomly-selected node in their NBRS lists. Adaptive Probabilistic Search(APS)[5]is similar to1-RW ALK.However,in APS,each node maintains a success ratio of previous searches for each object and uses the suc-cess ratio to select nodes.In our simulation,since the ob-jects are distributed randomly and the number of the objects is much larger than the number of queries,we expect that few queries will search for the same objects,and therefore, we believe that APS will behave similarly to1-RW ALK. We measure the performance in two primary metrics:the number of messages generated,and the number of hops per query.The number of messages is the total number of all messages generated by the involved nodes until the target node is found.The number of hops for a query is the num-ber of nodes on the path from the query initiator to the target node.By assuming that forwarding a query to other nodes takes roughly the same amount of time,the number of hops for a query is a good indication of the amount of time that the query took to complete.
We tested our algorithm with1000nodes with1600 queries.In both k-RW ALK and1-RW ALK,the k value is static and set to3.This is simply because with a k value larger than3,k-RW ALK requires too many messages to complete the simulation in reasonable time.In our ABS al-gorithm,k-min is set to3,k-threshold varies from0.4to0.8, and the k value(denoted the k d value)is dynamically deter-mined based on the k-min and k-threshold values.ABS-0.4,ABS-0.6and ABS-0.8are used to denote the ABS al-gorithm with k-threshold values0.4,0.6,and0.8,respec-tively.
In all three algorithms,each node N i maintains NBRS i, a set of nodes that N i knows.In k-RW ALK and1-RW ALK, NBRS i is a list of nodes,and as each node N i encoun-ters new nodes,the nodes are added to NBRS i.In ABS, NBRS i is a list of tuples,and only the tuple of a target node is added in the NBRS lists of the nodes on the path from the initiator to the target nodes.Thus,ABS’s NBRS lists do not grow as fast as those of k-RW ALK or1-RW ALK. In the simulation,the size of the NBRS lists is not lim-ited.
Nodes were created and linked randomly.For each link, its bandwidth,delay,and the queue handling mechanism are also randomly determined.Object placement on nodes is also done randomly with each nodes being assigned10ob-jects in average.Thus,each node may or may not have any objects.
4.1.Performance Comparison
Figure2(a)compares the number of messages required to complete searches,and Figure2(b)compares the num-ber of hops from an initiator to a target node.Also note that the average number of nodes that each node concur-rently forwards a query to is not always k d(in ABS),3(in k-RW ALK),or1(in1-RW ALK).This is because as a search progresses,eventually some nodes will claim a search to be “fail”;at this moment,the number of nodes that this node forwards the query to is0.We use k r to denote the actual number of nodes that a node forwards a query to.We use no-tation#QUERY to denote the number of queries that have been completed.
https://www.wendangku.net/doc/4c18191953.html,parison of k-RW ALK and1-RW ALK First, we compare the performance of k-RW ALK and1-RW ALK. We expect that if each node forwards more query messages at once,the total number of messages would increase,how-ever the hops required to complete a search would decrease. In addition,if the number of search paths increases,the probability of the same node being visited more than once
M e s s a g e s / Q u e r y
Queries
(a)Total messages generated.H o p s / Q u e r y
Queries
(b)Number of hops per query.
Figure 2.ABS performance under NS2for 1000nodes.
will increase (as discussed in Section 3).This would result in increase in the number of wasted messages.However,it would contribute to shortening search paths that claim “fail”.This is because when a node is visited by a query for the second time and onward,the node immediately re-turns “fail”to its sender.
When k-RW ALK and 1-RW ALK are compared,k-RW ALK requires signi?cantly more messages but much fewer hops than 1-RW ALK.In fact,1-RW ALK con-stantly requires an extremely small number of mes-sages,however a large number of hops.In contrast,even though k-RW ALK requires a constantly low num-ber of hops,its number of messages increases sharply.The reason for this is because the size of NBRS lists in-creases as the execution progresses.Then,the number of nodes that each node must explore before it claims “fail”will increase.This will result in the sharp increase in the number of required messages as NBRS lists expands,un-til the lists become large enough so that a search would terminate with “success”before all nodes in the lists are ex-plored.Note that as #QUERY increases in k-RW ALK the number of messages increases sharply,but the num-ber of hops does not change much.
Next,we analyze why this phenomenon is not clearly observed in 1-RW ALK.It is assumed that since the total number of search paths is limited to k ,the possibility of a node being visited multiple times is low.This will make each search path long;that is,each search path tends to visit many nodes sequentially before it claims “success”or “fail”.Thus,in 1-RW ALK,it is assumed that initial NBRS lists are already large and a search would terminate before all nodes in NBRS have been explored.Therefore,the fur-ther increase in the NBRS lists would have little effect on
the number of messages.
In the performance of k-RW ALK,the number of hops is stabilized at around 8after #QUERY reaches 75.The number of messages sharply increases until it stabilizes at around 320000when #QUERY reaches around 700.At this point,it is assumed that NBRS lists become so large that virtually no node claims “fail”before a search terminates,or that most NBRS lists cover all the nodes in the sys-tem and cannot expand any further.In the performance of 1-RW ALK,after #QUERY reaches somewhere around 10and 20,both the number of hops and the number of mes-sages are stabilized at around 170(very large)and 3000(very few),respectively.
4.1.2.Performance analysis of ABS Based on the above observation,we now analyze the performance and behav-ior of ABS.In both metrics,ABS demonstrates extremely good performance.
First,we compare ABS with 1-RW ALK.Even when the average k r values of ABS is small (up to 1.5when #QUERY is 9),there is a signi?cant difference in the number of hops:about 30(ABS)vs.170.This is because in ABS,even when the average k r value is small,k r would vary from 0to a value larger than 1,and as a result,there are many con-current search paths.In 1-RW ALK,individual k r values of nodes other than the initializing node N I are either 0or 1,and at most three search paths exist concurrently.The num-ber of hops in ABS stabilizes at k r value of around 8.The number of messages in ABS gradually increases to 15000,whereas 1-RW ALK stabilizes at around 3000.
Next we compare ABS with k-RW ALK.Up to #QUERY being around 200,k-RW ALK shows slightly better perfor-mance than ABS in terms of the number of hops.After that,ABS demonstrates even slightly fewer number of hops than
124810501003008001600 ABS0.720.89 1.29 1.45 1.54 2.42 4.2425.8550.9882.31 APS0.530.630.740.850.870.970.98 1.00 1.00 1.00 RW ALK0.82 1.22 1.8 2.37 2.46 2.89 2.94 2.98 2.99
Table1.k-values as the number of requests increases.
k-RW ALK.Importantly,however,ABS achieves such per-formance with a signi?cantly fewer(80%at18hops)mes-sages than k-RW ALK.
Among the ABSs,a smaller k-threshold value results in a larger k d value.As expected,ABSs with smaller k-threshold values require more messages than those with larger k-threefold values.However,no signi?cant differ-ence is found in the number of hops among ABSs with var-ious k-threshold values.
5.Conclusions and future work
In this paper we have presented a scalable P2P search al-gorithm using a method to dynamically determine the value of k d,the number of nodes to send a query to at once.The decision is based on the degree to which each neighbor has contributed to a successful search,called the activity level. Each node sends a query to K d neighbors with the high-est activity levels at once.A distinctive feature of the algo-rithm is that,unlike APS,a node does not consider the se-mantic preference when choosing the nodes to forward a query to.A node simply considers the previous success ra-tio,regardless of the objects being searched,to determine k d and the nodes to send a query to.Therefore,its scalabil-ity is better than that of APS in terms of the storage size. Our algorithm automatically creates a spanning graph consisting of high traf?c links.Once a stable spanning graph is created,the number of hops required for a search is roughly bound by the diameter of the spanning graph.The simulation exhibits that our algorithm demonstrates signif-icantly better performance than both k-random walk and 1-random walk.Our message-per-query is much less than k-random walk,yet our number of hops to reach the tar-get nodes is much less than1-random walk and slightly less than even k-RW ALK.
Our algorithm has signi?cant potential for further im-provement.For example,as APS does,we could re?ect the success ratio of each object or of each category of objects in evaluation of activity levels.Another possibility is to con-sider the capacities of nodes and links as[15]does when determining the activity levels.
Acknowledgments
This material is based in part upon work supported by the Na-tional Science Foundation under the award number CCR-0082667.Any opinions,?ndings,and conclusions or recommendations ex-pressed in this publication are those of the author(s)and do not necessarily re?ect the views of the National Science Foundation. References
[1]Gnutella website:https://www.wendangku.net/doc/4c18191953.html,.
[2]Q.Lv,P.Cao,E.Cohen,K.Li,and S.Shenker.Search and
replication in unstructured peer-to-peer networks.In Proc.of ICS,2002.
[3]S.Androutsellis-Theotokis.A Survey of Peer-to-Peer File
Sharing Technologies.White paper WHP-2002-03,ELTRUN, 2002.
[4]C.Gkantsidis,M.Mihail,and A.Saberi.Random walks in
peer-to-peer networks.In Proc.of IEEE INFOCOM,2004.
[5]Dimitrios Tsoumakos and Nick Roussopoulos.Adaptive
Probabilistic Search(APS)for Peer-to-Peer Networks.In IEEE Intl.Conf.on P2P Computing,2003.
[6]S.Ratnasamy,P.Francis,M.Handley,and R.Karp.A scalable
content-addressable network.In Proc.of SIGCOMM,2001.
[7]B.Y.Zhao,J.Kubiatowicz,and A.Joseph.Tapestry:An in-
frastructure for fault-tolerant wide-area location and routing.
Technical Report UCB/CSD-01-1141,Computer Science Di-vision,University of California,Berkeley,April2001. [8]Ronaldo A.Ferreira,Murali Krishna Ramanathan,Suresh Ja-
gannathan,and Ananth Grama.Ef?cient Randomized Search Algorithms in Unstructured Peer-to-Peer Networks.Purdue University Technical Report PUTR-CS04-022,2004.
[9]Napster website:https://www.wendangku.net/doc/4c18191953.html,.
[10]Kunwadee Sripanidkulchai,Bruc Maggs,and Hui Zhang.Ef-
?cient content location using interest based locality in peer-to-peer system.In Proc.of IEEE INFOCOM,2003.
[11]Alexander L¨o ser and Christoph Tempich.On Ranking Peers
in Semantic Overlay Networks.In3rd Conference on Profes-sional Knowledge Management,2005.
[12]Hailong Cai and Jun Wang.Foreseer:A Novel,Locality-
Aware Peer-to-Peer System Architecture for Keyword Searches.In Middleware2004,ACM/IFIP/USENIX Interna-tional Middleware Conference,2004.
[13]Alexander L¨o ser and Steffen Staab and Christoph Tempich.
Semantic Methods for P2P Query Routing.In3rd.German Conference on Multi Agent Technologies,2005.
[14]V.Cholvi,P.A.Felber,and E.W.Biersack.Ef?cient search
in unstructured peer-to-peer networks.Technical Report RR-03-090,Institut Eur`e com,2003.
[15]Yatin Chawathe,Sylvia Ratnasamy,Lee Breslau,Nick Lan-
ham and Scott Shenker.Making Gnutella-like P2P Systems Scalable.In Proc.of ACM SIGCOMM,2003.
[16]Network Simulator2https://www.wendangku.net/doc/4c18191953.html,/nsnam/ns/
中外金属材料对照表
常用国内外钢材牌号对照表 中国 美国 日本 德国 英国 法国 前苏联 国际标准化组织 GB AST JIS DIN 、DINEN BS 、BSEN NF 、NFEN ΓOCT ISO 630 品 名 牌号 牌号 牌号 牌号 牌号 牌号 牌号 Q195 Cr.B Cr.C SS330 SPHC SPHD S185 040 A10 S185 S185 CT1K П CTlC П CTl ПC Q215A Cr.C Cr.58 SS 330 SPHC 040 A12 CT2K П—2 CT2C П—2 CT2ПC —2 Q235A Cr.D SS400 SM400A 080A15 CT3K П—2 CT3C П—2 CT3ПC —2 E235B Q235B Cr.D SS400 SM400A S235JR S235JRGl S235JRG2 S235JR S235JRGl S235JRG2 S235JR S235JRGl S235JRG2 CT3K П—3 CT3C П—3 CT3ПC —3 E235B Q255A SS400 SM400A CT4K П—2 CT4C П—2 CT4ПC —2 普 通 碳 素 结 构 钢 Q275 SS490 CT5C П—2 CT5ПC —2 E275A
中国 美国 日本 德国 英国 法国 前苏联 国际标准化组织 GB AST JIS DIN 、DINEN BS 、BSEN NF 、NFEN ΓOCT IS0 630 品 名 牌号 牌号 牌号 牌号 牌号 牌号 牌号 08F 1008 1010 SPHD SPHE 040A10 80K П 10 1010 S10C S12C CKl0 040A12 XCl0 10 C101 15 1015 S15C S17C CKl5 Fe360B 08M15 XCl2 Fe306B 15 C15E4 20 1020 S20C S22C C22 IC22 C22 20 25 1025 S25C S28C C25 IC25 C25 25 C25E4 40 1040 S40C S43C C40 IC40 080M40 C40 40 C40E4 45 1045 S45C S48C C45 IC45 080A47 C45 45 C45E4 50 1050 S50C S53C C50 IC50 080M50 C50 50 C50E4 优 质 碳 素 结 构 钢 15Mn 1019 080A15 15r
不锈钢材料牌号对照表
0Cr18Ni9作为不锈钢耐热钢使用最广泛，用于食品用设备，一般化工设备，原子能用工业设备。通俗的讲0Cr18Ni9就是304不锈钢板，0Cr18Ni9Ti就是321，一个是国标，一个是美标。321是因为原来冶炼技术不好，无法降低碳含量才研制的，现在因冶炼技术的提高，超低碳钢冶炼已经很平常，所以321有被淘汰的趋势。目前321的产量已经很少了。只有一些军工还在使用。0Cr18Ni9钢(AISI304)是奥氏体不锈钢，是在最初发明的18-8型奥氏体不锈钢的基础上发展演变的钢种，该钢是不锈钢的主体钢种，其产量约占不锈钢总产量曲30%以上。由于此钢具有奥氏体结构，它不可能通过热处理手段予以强化，只能采用冷变形方式达到提高强度的目的。钢的奥氏体结构赋予了它的良好冷、热加工性能、无磁性和好的低温性能。0Cr18Ni9钢薄截面尺寸的焊接件具有足够的耐晶间腐蚀能力，在氧化性酸(HNO3)中具有优良的耐蚀性，在碱溶液和大部分有机酸和无机酸中以及大气、水、蒸汽中耐蚀性亦佳。 0Cr18Ni9钢的良好性能，使其成为应用量最大、使用范围最广的不锈钢牌号，此钢适于制造深冲成型的部件以及输送腐蚀介质管道、容器，结构件等，0Cr18Ni9亦可用子制造无磁、低温设备和部件。 0Cr19Ni10(AISI304L)是在0Cr18Ni9基础上，通过降低碳和稍许提高含镍量的超低碳型奥氏体不锈钢。此钢是为了解决因Cr23C6析出致使0Cr18Ni9钢在一些条件下存在严重的晶间腐蚀倾向而发展的。在开发初期，因冶金生产降碳较难，一度曾妨碍了它的广泛应用，在20世纪70年代新的二次精炼方法AOD和VOD工艺成功用于生产后，此钢才真正得到广泛应用。与0Cr18Ni9比较，此钢强度稍低，但其敏化态耐晶间腐蚀能力显著优于0Cr18Ni9。除强度外，此钢的其他性能同于0Cr18Ni9。它主要用于需焊接且焊后又不能进行面溶处理的耐蚀设备和部件。上述两个钢种，在易产生应力腐蚀环境和产生点蚀和缝隙腐蚀的条件下，在选用时应慎重。[1] 特性 具有良好的耐蚀性、耐热性、低温强度和机械性能，冲压弯曲等热加工性好，无热处理硬化现象，无磁性。 用途 家庭用品、橱柜、室内管线、热水器、锅炉、浴缸、汽车配件、医疗器具、建材、化学、食品工业、农业、船舶部件。 化学成份
公制,英制,美制螺纹尺寸对照表
螺纹简介：
1.1螺纹的起源 一般认为，阿基米德（公元前287-212年）是首先将螺纹用于工业目的，即将低处的水移到高处的一种工具。1659年法国人贝兹逊发明螺纹切割机；其后，经过英国的威阿特、亨利?莫斯列，到其子弟约瑟夫?惠氏于1841年才统一原本混乱的螺纹，促进螺纹制品的全球普及；因此，惠氏螺纹也随英国工业的发达而广泛传播世界。 美制螺纹依据塞勒氏螺纹（Seller's thread）而问世于1924年；二次世界大战中，美国、英国、加拿大三国协定发展成统一制螺纹Unified thread。随着国际交流的频繁，1947年成立的ISO国际标准化组织（International standard organization)成为推动国际标准化的重要力量。 1.2螺纹概述 按用途分类，可分为紧固螺纹、密封螺纹、传动螺纹、管螺纹、专用螺纹。 1.3螺纹的分类 A、公制螺纹 表示螺纹的记号为M,公称直径d或D与螺距P是以mm毫米来标识，螺纹牙尖角为60度。这是我国GB标准所规定使用的螺纹，也是常见到的。 B、英美制统一螺纹 表示螺纹的记号为U,公称直径d或D为英寸，螺距是以每英寸（25.4mm）的牙树来标识，螺纹牙尖角为60度。这是通俗成为美制规格的螺纹。 C、惠氏螺纹 表示螺纹的记号为W，螺纹直径为英寸，螺距以每英寸（25.4mm）的牙树来标识，螺纹牙尖角为55度。这是英国采用的标准，牙尖角有点特殊的螺纹。 D、管用螺纹 主要指英国BSP螺纹系统，用在管接头，分为推拔螺纹（记号为PT）和平和螺纹（记号为PS），牙尖角为55度，螺距以每英寸（25.4mm）的牙树来标识；公称方法依据所配的钢管的公称方法（英寸），因此，实际的螺纹外径与公称直径不同。 E、自行车用螺纹
统一螺纹标准(美制螺纹)
统一螺纹标准（美制螺纹标准） 统一螺纹标准（UTS）规定了一种标准螺纹形式和其他一系列数据，包括公差、容差和名称，通常多用于美国和加拿大。它和世界其他地方常用的ISO公制螺纹相同，有着60°的牙型角，只是UTS螺纹的主要尺寸（外径和螺距）以英寸为单位，用分数表示，而不是以毫米为单位的整数值。UTS目前由美国机械工程师协会/美国国家标准化组织（ASME/ANSI）管理控制。 基本齿廓 在本系列螺纹中，每个螺纹由大径Dmaj和螺距P规定。UTS螺纹包括一个对称的V形螺纹。在螺纹轴线 的平面上，V形两边形成60° 的夹角，V形高度为H，其最上 的0.125和最下的0.25被切 掉。 螺距P指两个螺纹顶端之 间的距离。UTS螺纹为单线螺 纹，螺距等于导程（螺纹每 360°的轴向距离）。UTS螺纹通 常不用螺距这一参数，而用每 英寸的牙数（TPI），它与螺距 互为倒数。高度H和螺距P关系如下： 在外螺纹中（如螺栓），大直径Dmaj和小直径Dmin规定了螺纹的最大尺寸。这也就是说，外螺纹直径最大为Dmaj，但最小直径可小于小径Dmin。相反，在内螺纹中（如螺母），大径和小径为最小尺寸，因此螺纹最小尺寸为Dmin,但最大尺寸可大于Dmaj。 小径Dmin和有效中径Dp是通过大径和螺距得来的，见下式： 名称 UTS螺纹的标准名称由一个表示螺纹公称直径（大径）的数字，后跟每英寸的牙数表示。对直径小于1/4英寸的，标准中规定用一个整数表示。所有其他直
径均以英寸表示。 这组数字后可跟字母UNC、UNF或UNEF，分别表示“粗螺纹”、“细螺纹”或“超细螺纹”，或者也可跟公差级别。 例： 6-32 UNC 2B （大径：0.1380英寸，螺距：32牙/英寸） 常用尺寸 这些数字大致遵循一个对数级数，每增加一个螺纹尺寸，其螺纹抗拉强度约增加一倍。 规格大于等于0#的螺纹，可用下列公式计算其大径：大径=螺纹#×0.013＂
世界各国不锈钢牌号对照表
SUS304L与SUS304特性及化学成分含碳量不一样。 SUS304L具备SUS304所有特性；耐腐蚀性和耐热性优良；焊接后或消除应力后，亦能保持良好耐腐蚀性；深冲性，抛光性优良。应用于抗晶间耐腐蚀性高的化学、煤炭、石油环境中的器械和设备；露天装饰材料、耐热零件及热处理有困难的零件。SUS304L含碳量≤0.03 SUS304是奥氏体基本钢种，用途最为广泛.耐腐蚀性和耐热性优良；低温强度和机械性能优良；单相奥氏体组织，无热处理硬化现象(无磁性，使用温度-196至-800℃)。SUS304含碳≤0.08 东莞伟鑫不锈钢材料有限公司是一家批发兼零售的企业公司，专业经营各种进口、国产一级品牌的精密环保不锈钢带（卷带、卷料、带材）、不锈钢棒（六角棒、方棒、易车棒、研磨棒）、不锈钢线（螺丝线、弹簧线、车轴线、全软线）、不锈钢管（无缝管、装饰管、工业管）、不锈钢板（不锈钢中厚板、不锈钢卷板、不锈钢8K板） 本公司始终秉承“高品质满足高要求”的经营宗旨，坚持“完美品质值得信赖”的企业发展理念同时以快捷、高效和完善的售后服务结合与客户共命运的“供应链式”营销，以专业的角度，协助客户选择最合适的不锈钢材料，有效的降低了生产成本，提高了市场竞争力。赢得了广大用户的赞誉，建立了长期稳定的不锈钢材料供应关系。产品符合中国GB，美国ASTM、日本JIS等规范，所有出厂产品都有RoHS指令的SGS报告和材质证明。 主要品牌有：新日铁、JFE、住友、宝钢、浦项、象牌、日钢、太钢、张浦、联众、克虏伯等。 不锈钢带材 规格：宽度4mm～1550mm 厚度0.025mm～4mm 材质：SUS201、202、301、304、304L、316、316L、310、310S、420、430 表面：2B、BA、8K镜面，拉丝 不锈钢棒材 规格：Ф2mm～Ф280mm 材质：SUS202、301、303、303F、304、304L、316、316L、310、310S、321 不锈钢线材 规格：Ф0.02mm～Ф12mm 材质：SUS202、304、304HC、302、302HQ、316、316L、316Ti、310、310S、420、430等 不锈钢六角棒 规格：H2mm～H90mm 材质：SUS202、303、303F、304、304L、316、316F、316L、310、310S、321 不锈钢板材 规格：0.4mm～80mm 材质：SUS201、202、301、303、304、304L、316、316L等。 不锈钢管材 规格：Ф2mm～Ф426mm×1.0～50.0 材质：201、202、301、302、304、304L、316、316L、310S、310 不锈钢钢丝绳 规格：Ф0.15mm-Ф46mm 1 7、7 7、1 19、6 19、7 19、6 37、7 37等。 材质：SUS316、316L、304、304L、303、303Cu、302、301、202、201等，光面绳；雾面绳；硬绳；软绳；PC；PE；PVC包胶绳等（且承接定做特殊规格的不锈钢绳）。
公制美制螺纹牙距表
1,公制螺纹牙距表 公称直径牙距mm 毫米粗牙细牙 M1 0.25 0.2 M1.2 0.25 0.2 M1.6 0.35 0.2 M2 0.4 0.25 M2.5 0.45 0.35 M3 0.5 0.35 M4 0.7 0.5 M5 0.8 0.5 M6 1 0.75 M8 1.25 1 M10 1.5 1.25 M12 1.75 1.25 M14 2 1.5 M16 2 1.5 M18 2.5 1.5 M20 2.5 1.5 M22 2.5 1.5 M24 3 2 M27 3 2 M30 3.5 2 (M33) 3.5 2 M36 4 3 (M39) 4 3 M42 4.5 3 (M45) 4.5 3 M48 5 3 2,美制统一标准螺纹牙数表 公称直径直径尺寸牙距 in mm 粗牙细牙 0# 0.06 1.524 - 80 1# 0.073 1.854 64 72 2# 0.086 2.184 56 64 3# 0.099 2.515 48 56 4# 0.112 2.845 40 48 5# 0.125 3.175 40 44 6# 0.138 3.505 32 40 8# 0.164 4.166 32 36 10# 0.19 4.826 24 32 12# 0.216 5.486 24 28
1/4" 0.25 6.35 20 28 5/16" 0.3125 7.938 18 24 3/8" 0.375 9.525 16 24 7/16" 0.4375 11.113 14 20 1/2" 0.5 12.7 13 20 9/16" 0.5625 14.288 12 18 5/8" 0.625 15.875 11 18 3/4" 0.75 19.05 10 16 7/8" 0.875 22.225 9 14 1" 1 25.4 8 12 1-1/8" 1.125 28.575 7 12 1-1/4" 1.25 31.75 7 12 1-3/8" 1.375 34.925 6 12 1-1/2" 1.5 38.1 6 12 1-3/4" 1.75 44.45 5 - 2" 2 50.8 4.5 – 6-32是每英寸牙数(在25.4mm长度上的牙数) 螺距=25.4/每英寸牙数 例如：6牙 螺距=25.4/6=4.2333
各国常用材料牌号对照表
材料类型应用 承 (内 )压铸件铸铁 一般铸件 承 (内 )压铸件 轧材 /锻件 棒材 :承(内) 压 棒材 :一般碳钢 螺栓和螺柱 螺母 板材 管材 各国常用材料牌号对照表 材料牌号 中国德国美国 GB DIN ASTM UNS HT2500.6025GG-25 A 278 Class 3.0 F 12401 HT3000.6030GG-30 A 48 Class25/30 F 11701/F 12101 ZG230-450 1.1155GS-Ck25 A 216 Gr LC8J 03 003 Q235 1.0402C22A266 Class2K 03506 Q235 1.0402C22 A 295 Gr D40G 10 200 45 1.0503C45A576 Gr 1045G10 450 30CrMo 1.725824CrMo5 A 193 Gr B7G41 400 35 1.1181CK35 A 194 Gr 2H K 04 002 Q255 1.0254St.37.0 A 516 Gr 65/70K 02 403/K 02 700 20Mn 1.0305St.35.8 A 106 Gr B K 03 006
材料类型AISI410 钢12%铬钢 材料牌号 应用中国德国美国 GB DIN ASTM UNS 配件20 1.0308St35.0 A 105K 03 504棒料42CrMo 1.722542CrMo4 A 434 Class BB G 41 400 c 螺栓和螺柱35CrMoV 1.771140CrMoV47 A 193 GrB7G 41 400螺母30CrMo 1.725824CrMo5 A 194 Gr 2H K 04 002 ZG1Cr13 1.4008G-X8CrNi13 A 217 Gr CA 15J 91 150承 (内 )压铸件 1.4313G-X5Ni13 4 A 487 Gr CA6NM J 91 540 轧材 /锻件 : 1Cr13 1.4006X10Cr13 A 182 Gr F6a Class1S 41 000 承(内)压 1.4313X4CrNi13 4 A 182 Gr F6NM S 41 500轧材 /锻件 :一般 1.4313X4CrNi13 4 A 473 Type 410S 41 000棒料 : 承(内 )压1Cr13 1.4006X10Cr13 A 479 Type 410S 41 000棒料 :一般 A 276 Type 410S 41 000
国内外常用材料牌号对照
国内外常用材料牌号对照(仅供参考) 德国 国际标准 美国 英国 法国 日本 中国 DIN ISO AISI UNS ASTM BSI AFNOR JIS GB 17007 17006 序号 材料 类型材料号 牌号 标准牌号 牌号 标准号 牌号 牌号 牌号 牌号 牌号 1 0.6018 GG18（GG15?） A48 Cl 25 1452Gr180 FC15 HT150 2 0.6020 GG20 185Gr 200 A48 Cl 30 1452Gr220 Ft20 FC20 HT200 3 0.6022 GG22 A48 Cl 30 1452Gr220 FC20 HT200 4 .06025 GG25 185Gr 250 A48 Cl 35,45 1452Gr260 Ft25 FC25 HT250 5 .06030 GG30 185Gr 300 A48 Cl 45 1452Gr300 Ft30 FC30 HT300 6 0.7040 GGG-40 1083/400-12 A536 60-40-18 2789Gr370/17 2789Gr420/12 FGS 400-12 FCD40 QT40-17 7 0.7042 GGG-42 1083/400-12 A536 60-40-18 2789Gr420/12 FGS 400-12 FCD40 QT40-17 8 铸 铁0.7050 GGG-50 1083/500-7 A536 60-45-12 2789Gr500/7 FGS500-7 FCD45 QT42-10 9 1.0420 GS-38.3 A27 Gr60-30 3100GrA2 230-400M SCA1 ZG15 10 1.0443 GS-45 A27 Gr65-35 3100GrA3 280-480M SCA2 ZG25 11 铸 钢 1.0446 GS-45.3 A27 Gr70-36 3100GrA3 280-480M SCA2 ZG25 12 1.4008 G-X12Cr14 A743 GrCA-15 3100 410C21 Z12CN13.02M SCS1 ZG1Cr13 13 1.4027 G-X20Cr14 A743 GrCA-15 GrCA-40 3100 420C29 (Z20C13M) SCS2 ZG2Cr13 14 1.4059 G-X22CrNi17 A743 GrCB-30 3146ANC2 Z22CN17.01M SCS12 ZG1Cr17Ni2 15 1.4138 G-X120CrMo292 A743 ～GrCC-50 Z120CD29.02 M 16 1.4308 G-X6CrNi189 J92600 J92650 A351 A743 GrCF-8 1504 304C15 3100 304C15 Z6CND18.10M SCS13 ZGoCr18Ni9 17 1.4313 g-x5cRnI134 A743 A757 GrCA-6NM GrE3N Z6CN13.04M SCS5 18 合金铸 钢 1.4336 G-X20CrNi248 (Z20CN24.08 M)
世界材料牌号对照表
世界材料牌号对照表 材料对照表工具钢 国家和工标准件加工 材特性德国中国英国法国意大利比利时瑞典西班牙日本美国中国料组材料编DIN GB BS EN AFNOR UNI NBN SS UNE JIS AISI/SAE GB 组号6/7 1.1545 C105W1 T10 - - Y1105 C98KU;C100KU - 1880 F.515;F.516 - W.110 T10 6/7 1.1663 C125W T12A - - Y2120 C120KU - - (C120) SK2 W.112 T12A 9 1.2067 100Cr6 CrV;9SiCr BL3 - Y100C6 - - - 100Cr6 - L3 CrV;9SiCr X210Cr13KU; 10/11 1.2080 X210Cr12 Cr12 BD3 - Z200Cr12 - - X210Cr12 SKD1 D3 Cr12 X250Cr12KU X35CrMoV05KU; 10/11 1.2344 X40CrMoV51 4Cr5MoVSi BH13 - Z40CDV5 - 2242 X40CrMoV5 SKD61 H13 4Cr5MoVSi X40CrMoV51KU 10/11 1.2363 X100CrMoV51 Cr6WV BA2 - Z100CDV5 X100CrMoV51KU - 2260 X100CrMoV5 SKD12 A2 Cr6WV 10WCr6; SKS31; 10/11 1.2419 105WCr6 CrWMo - - 105WC13 - 2140 105WCr5 - CrWMo 107WCr5KU SKS2;SKS3 10/11 1.2436 X210CrW12 Cr12W - - - X215CrW121KU - 2312 X210CrW12 SKD2 - Cr12W 10/11 1.2542 45WCrV7 5CrNiMo BS1 - - 45WCrV8KU - 2710 45WCrSi8 - S1 5CrNiMo X30WCrV93; X28W09KU 10/11 1.2581 3Cr2W8V BH21 - Z30WCV9 - - - - - 3Cr2W8V X30WCrV93KU X30WCrV93KU P
常用材料牌号对照
类别标准号常用钢号相应ASTM材料 1 螺栓GB5782－86 GB5785－86 SH3404－96 HG20613－97 HG20634－97 Q235B 35 0Cr18Ni9 A307-B A325-I A193-B8、A320-B8 2 螺柱GB901－88 SH3404－96 HG20613－97 HG20634－97 35 35CrMoA 25Cr2MoVA 0Cr18Ni9 A325-I A193-B7、A320-L7 A193-B16 A193-B8、A320-B8 3 螺母GB6170－86 GB6171－86 GB6175－86 GB6176－86 SH3404－96 HG20613－97 HG20634－97 45 30CrMo 0Cr18Ni9 A194-2H A194-4、A194-7 A194-8 1 对焊管件 管件类别标准号常用钢号相应美国标准相应ASTM材料1.1 无缝对焊管件 （包括弯头、三通、四通、异径管、管帽、翻边短节）GB12459－90 HGJ514－87 SH3408－96 20 16Mn 15CrMo 12Cr1MoV 1Cr5Mo 0Cr19Ni9 00Cr19Ni10 0Cr17Ni12Mo2 00Cr17Ni14Mo2 ANSI B16.9 ANSI B16.28 MSS SP-43 A234-WPB A420-WPL6 A234-WP12 A234-WP11 A234-WP5 A403-WP304 A403-WP304L A403-WP316 A403-WP316L
1.2 焊缝对焊管件 包括（弯头、三通、四通、异径管、管帽）GB/T13401－92 HGJ528－90 SH3409－96 Q235A、20g 16MnR 0Cr19Ni9 00Cr19Ni10 0Cr17Ni12Mo2 00Cr17Ni14Mo2 ANSI B16.9 ANSI B16.28 A234-WPB A420-WPL6 A403-WP304 A403-WP304L A403-WP316 A403-WP316L 2.锻制管件 管件类别标准号常用钢号相应美国标准相应ASTM材料2.1 承插焊管件 包括（弯头、三通、四通管接头、活接头、支管台、管帽）GB/T14383－93 HGJ10－88 HGJ529－90 SH3410－96 20 16Mn 15CrMo 12Cr1MoV 1Cr5Mo 0Cr19Ni9 00Cr19Ni10 0Cr17Ni12Mo2 00Cr17Ni14Mo2 ANSI B16.11 MSS SP-83 MSS SP-97 A105 A350-LF2 A182-F12 A182-F11 A182-F5 A182-F304 A182-F304L A182-F316 A182-F316L 2.2 螺纹管件 包括（弯头、三通、 四通管接头、活接头、支管台、管帽、丝堵、内外螺纹接头）GB/T14626－93 HGJ529－90 20 16Mn 15CrMo 12Cr1MoV 1Cr5Mo 0Cr19Ni9 00Cr19Ni10 0Cr17Ni12Mo2 00Cr17Ni14Mo2 ANSI B16.11 MSS SP-83 MSS SP-97 A105 A350-LF2 A182-F12 A182-F11 A182-F5 A182-F304 A182-F304L A182-F316 A182-F316L
常见材料中外牌号对照
种类中国GB 国际标准ISO 前苏联гOCT德国DIN 美国ASTM UNS AISI 日本JIS 英国BS 法国NF 碳素结构钢Q195(A1) ＃Cт/сп＃A283grA ＃＃SS34 040A10 A33 碳素结构钢Q195-b(A1 b) Cт/пс 碳素结构钢Q195-F(A1F) Cт/кп 碳素结构钢Q215-A(A2) Cт2псRSt34-2 A283grB SS34 040A12 A34-2 碳素结构钢Q215-A.b(A2b) Cт2сп 碳素结构钢Q215-A.F(A2F) Cт2кпUSt34-2 碳素结构钢Q215-B(C2) BCт2спRSt34-2 A283grB 040A12 A34-2 碳素结构钢Q215-B.b(C2b) BCт2пс 碳素结构钢Q215-B.F(C2F) BCт2кпUSt34-2 碳素结构钢Q235-A(A3) Cт3спRSt37-2 A283grC SS41 050A17 A37-2 碳素结构钢Q235-A.b(A3b) Cт3пс 碳素结构钢Q235-A.F(A3F) Cт3кпUSt37-2 碳素结构钢Q235-B(C3) BCт3спRSt37-2 A36 SS41 050A17 A37-2 碳素结构钢Q235-B.b(C3b) BCт3пс 碳素结构钢Q235-B.F(C3F) BCт3кпUSt37-2 碳素结构钢Q235-C 060A22 A42-2 碳素结构钢Q235-D 060A22 A42-2 碳素结构钢Q255-A(A4) Cт4спRSt42-2 A238grD 060A22 A50-2 碳素结构钢Q255-B(C4) BCт4спRSt42-2 碳素结构钢Q275(C5) BCт5спSt50-2 SS50 优质碳素钢08F 08кп 优质碳素钢10F 10кп 优质碳素钢15F 15кп 优质碳素钢8 8 G1008 1008 S09CK 050A04 优质碳素钢10 C10e 10 CK10 G10100 1010 S10C 050A10 XC10 优质碳素钢15 C15e 15 CK15 G10150 1015 S15C 050A15 XC12 优质碳素钢20 C20e 20 CK20 G10200 1020 S20C 050A20 XC18 优质碳素钢25 C25e 25 CK25 G10250 1025 S25C 060A25 XC25 优质碳素钢30 C30e 30 CK30 G10300 1030 S30C 060A30 XC32 优质碳素钢35 C35e 35 CK35 G10350 1035 S35C 060A35 优质碳素钢40 C40e 40 CK40 G10400 1040 S40C 060A40 XC38H1 优质碳素钢45 C45e 45 CK45 G10450 1045 S45C 060A47 XC42H1 优质碳素钢50 C50e 50 CK50 G10500 1050 S50C 060A52 XC48H1 优质碳素钢55 C55e 55 CK55 G10550 1055 S55C 060A57 XC55H1 优质碳素钢60 C60e 60 CK60 G10600 1060 S58C 060A62 优质碳素钢65 65 CK67 G10650 1065 060A67 优质碳素钢70 70 G10700 1070 060A72 XC68
常用模具材料牌号对照表
常用模具材料牌号对照表 类别中国钢号通用钢号钢材特性 塑胶模具钢3Cr2Mo P20（美国）618(瑞典) 预硬塑胶模具钢 3Cr2NiMo 718（瑞典）P20+Ni(美国) 超预硬塑胶模具钢 4Cr13 S136（瑞典）抗腐蚀塑胶模具钢 1CrNi3 NAK80（日本）镜面塑胶模具钢 3Cr17Mo M300（奥地利）耐腐蚀塑胶模具钢 五金模具钢CrWMn SKS3（日本）不变形油钢 Cr12 Cr12MoV SKD11（日本）D3(美国) 耐磨韧性铬钢 Cr12Mo1V1 D2(美国) 热作模具钢4Cr5MoSiV1 SKD61（日本）通用热作模具钢 H13（美国） 8407（瑞典） 冷作模具钢 CrWMn/SKS31/105W/Cr6高硬度，中等淬透性，价格低廉。 207-255 820-840 下料模、冲头、成型模、搓丝板顶出杆及小型塑料压模等。 9Mn2V/O2/DF-2 具有良好冲载能力，热处理变形小。≤229 780-800 厚度小于6mm以下 的小型冲压模具及切纸机、刀具等。 9CrWMn/O1/SKS3/DF-3/100Mn/CrW4 淬火变形小，具有良好的刃口保持能力，热处理变形小。197-241 820-840 薄片冲压模、手饰压花模等。 9SiCr/X100Cr/MoV51具有高硬度良好的韧性和较好的抗回火稳定性。197-241 860-880 下料模、冲头、搓丝板、压印模、顶出杆等 Cr5Mo1V/A2/SKD12/XW-10/210/Cr12空冷淬硬性铬钢，韧性极佳，高耐磨损性和抗腐蚀能力。≤255 950-1000 拉伸模、压花模、下料模、冲压模、及耐磨塑料模等。 Cr12/D3/SKD1/X165Cr/MoV12高碳铬钢，具有高耐磨性和抗腐蚀能力。217-269 950-980 应用于小动载条件下要求高耐磨形状简单的拉伸模及冲载模。 Cr12MoV/X155Cr/VMo121具有良好的淬透性，高耐磨性，韧性高。 207-255 1000-1020 下料模、冲头、滚丝轮、剪刀片、冷镦模、陶土模及热固塑料成型模等。 Cr12Mo1V1/D2/SKD11/W-42具有良好的淬透性，高韧性，高耐磨损性，强韧性极佳，并具有良好的抗回火稳定性，热处理变形小。≤255 1000-1020 重型落料模、冷挤压模、深拉伸模、滚丝模、剪刀片、冷镦模、陶土模等。 7Cr7Mo2V2Si具有高韧性，高耐磨损性，热处理变形小。 241-269 1100-1150 适用于要求强韧性及高冲击载荷下工作的冷镦、冷冲等作业模具，特别适用于标准件和钢球
世界各国金属材料牌号对照汇总
以下是世界各国金属材料牌号对照汇总，仅供参考： Q195 = GradeB = SS330 = SPHC = S185 = 1.0035 Q215 = GradeC = CSTypeB = SS330 = SPHC Q235-A = GradeD = SS440 = S235JR = 1.0038 Q235-B = GradeD = SS440 = S235JO = 1.0114 Q235-C = GradeD = SS440 = S235J2 = 1.0117 Q235-D = SS440 = S235JR =1.0038 Q255-A = Grade 36[250] = SM400A = SS400 Q255-B = Grade 36[250] =SM400B Q275 = SS Grade 40[275] = SS490 = s275JR =1.0044 = S275JO = 1.0143 = S275J2 = 1.0145 Q295A = Grade 42[290] = SPFC 490 = E295 = 1.0050 Q295B = Grade 42[290] = SPFC 490 = E295 = 1.0050 Q345A = Grade 50[345] = SPFC 590 = E335 = 1.0060 Q345B = Grade 50[345] = SPFC 590 = S355JR = 1.0045 Q345C = Grade 50[345] = SPFC 590 = S355JO = 1.0553 Q345D = Grade 50[345] = SPFC 590 = S355J2 = 1.0577 Q345E = Grade 50[345] = SPFC 590 = S355NL = 1.0546 Q390A = Grade 55[380] = STKT 540 Q390B = Grade 55[380] = STKT 540 Q390C = Grade 55[380] = STKT 540 Q390D = Grade 55[380] Q390E = Grade 55[380]
常用材料牌号对照表 (1)
种类 GB国标ASTM美标JIS日标BS英标DIN德标 标准牌号标准牌号标准牌号 标准 牌号标准 牌号材料号 灰铸铁GB/T9439-1988 HT150 A48 No20A G5501 FC150 1452 150 1691 GG-15 0.6020 HT200 No25A G5501 FC200 1452 180 1691 GG-20 HT200 No30A G5501 FC200 1452 220 1691 GG-20 HT250 No35A - FC250 - 250 - GG-20 0.6025 HT300 No40A - FC300 - 300 - GG-30 0.6030 HT350 No50A - FC350 - 350 - GG-35 0.6035 GB/T9439 HT100 - - - FC-100 - - - GG-10 0.6010 球 墨铸铁GB/T9439-1988 QT400-18 A536 60-40-18 - FCD400-15 - 2789 420/12 GGG-40 0.7049 - A536 65-45-12 - FCD450-10 - 500/7 - GGG-50 0.7050 QT500-7 - - G5502 FCD500-7 2789 500/7 - - - QT600-3 A536 80-55-06 - FCD600-3 - 600/3 1693 GGG-60 0.7060 QT700-2 - 100-70-03 - FCD700-2 - 700/2 - GGG-70 0.7070 可焊 高温用碳素钢GB/T12229-1989 WCA级 A216 WCA G5151 SCPH1 1504 161Grade 430F 17245 GS-C25 1.0619 WCB级WCB SCPH2 161Grade 480E - - - WCC级WCC - - - - - ZG20Mo A217 WC1 G5151 SCPH-11 1504 245 17245 GS-22Mo4 1.5419 高温高压用合合金 - A217 WC4 G5151 SCPH-21 1504 621 17245 GS-14CrMo55 1.7357 ZG20CrMo WC5 - - 310 B2 - - - WC6 WC6 - - - - - - - WC9 WC9 G5151 SCPH-32 1504 622 - - - 1
世界各大公司功率材料牌号对照表[1]
世界各大公司功率材料牌号对照表 cross power material reference table DMEGC DMR30 DMR40 DMR44 R3K DMR50 DMR50B DMR55DMR24DMR95 TDK PC30 PC40 PC44 PC50 PC90 PC95 TDG TP3 TP4 TP4A TP5 TP5A ACME P4 P41 P51 P50 P46 TOKIN BH2 BH1 B40 BH5 N41N49(对应DMR50H) N92 EPCOS N27/N72 N67 N87/N97 3C92 FERROXCUBE 3C85 3C90 3C94/3C96 3F35 3F3 SAMWHA PL-5 PL-7 PL-11 PL-F1 PL-9 PM9 PM11 FM5 FM4 ISU PM5 PM7 MMG F5/F5A F44 F45 F47 NICERA NC-2H 2HM5 2M 5M ISKRA 45G 35G 55G 75G FDK 6H10/6H20 6H40 7H10 6H42 JFE MB3 MB4 MC2 MB1H MBT1 MAGNETICS P R K
世界各大公司高导材料牌号对照表 cross power material reference table DMR72DMR73R4K R5K R7K R10K R12K R15K DMEGC DMR71 H5C3 TDK DN45 DNW45H5A HP5 H5B2 H5C2 H5C4 TDG TH4 TL5 TL7 TL10 TH10 TS13 TS15 A043 A05 A07 A10 A101/A102A121 A15 A041 ACME N42 TOKIN 5H 7H 10H 12H 15H T57 N30 T35/T37 T38 T38 T42 T46 EPCOS N45 FERROXCUBE 3B46 3E28 3C113E27 3E5 3E55 3E6 3E7 SAMWHA ST-30B SM-35T SM-43T SM-50SM-70S SM-100 SM-150 ISU BM30 HM2A HM3 HM5A HM7 F39/FTA MMG F9 F10 FT7/F57 NICERA NC-5Y NC-7 NC-10H10TB 12H 15H FDK 2H042H06 2H07 2H10 2H5Z 2H15 MAGNETICS J W H STEWARD 36 46 35 42 40 45 JFE MA055MA070 MA100 MA101 MA120MA150 2H4A 2G4 2G1 2E2B 2H2A 2H1 TOMITA 2H5
中、美、日常用金属材料牌号对照[1]
看我国， 中、美、日常用金属材料牌号对照 公、英制螺纹对照 公、英制表面粗糙度、光洁度对应表 （内部参考） 上海一冷开利空调设备有限公司 技术部 2003年8月
前言 本资料所列的金属材料牌号对照，是以中国标准（GB）的牌号为基础，对照国际标准（ISO）和美国、日本的材料牌号，凡化学成分及主要机械性能相同或相似的材料，分类予以考虑，由于各国对材料成分的规定和牌号的命名与方法不尽相同，而且各国生产和冶炼条件和方法也有差异，很多材料相似但不完全相同，因此在材料选用时应根据产品实际需要，作出选择。 英制螺纹转换为公制（米制）螺纹时，依据最接近的公称直径来确定。另外还应考虑到螺距，根据实际需要，在转换时可以向上或向下一档转换。 各栏中括号内系旧牌号，现一并录入以供参考。 资料中出现的标准代号说明如下： AA ---- 美国铝学会标准 AISI ---- 美国钢铁学会标准 ASTM ---- 美国材料和试验协会标准 AWS ---- 美国焊接学会标准 GB ---- 中国国家标准 JIS ---- 日本工业标准 ISO ---- 国际标准
目录 1．中、美、日常用金属牌号对照 A．碳素结构钢（Carbon structural steel） B．锅炉和压力容器用钢（Carbon steel for boiler and pressure vessel）C．优质碳素钢（Quality carbon structural steel） D．合金结构钢杯（Alloy structural steel） E．碳素工具钢板（Carbon tool steel） F．不锈钢板（Stainless steel） G．耐热钢（Heat-resisting steel） H．铸铁（Cast iron） I．焊条（Welding rod） J．铜及铜合金（Copper and copper alloy） K．铝及铝合金（Aluminum and aluminum alloy） 2．公英制螺纹对照 A．公英制普通螺纹对照 B．用螺纹密封的管螺纹 C．非螺纹密封的管螺纹 3．公英制表面粗糙度和表面光洁度对应表