【正文】 t DRA can not be used in those algorithms.This paper presents an algorithm to determine the minimal amount and distribution in batches of DRA in pipelines with one input node and one output node (direct pumping). The algorithm is based on a bestfirst search algorithm, with the help of a scheduling purpose pipeline simulator and some approximate equations. The paper is organized as follows: Section 2 describes the problem and the tools used to solve the problem. Section 3 introduces the optimization algorithm. In section 4 some results on a real pipeline are presented to demonstrate the feasibility of the method. Finally, some concluding remark are given.Pipeline stateThe state of the pipeline is represented by the position of every batch inside the pipeline. Notice that knowing the pumped volume at the entry point, the state of the pipeline can be determined. As mentioned above, when using DRA we are considering maximum flow, then, during the pumping all allowable pumps are connected. As a consequence, the state’s flow is unique.Taking into account the dynamic behaviour of the system, the second key variable is time. Then, the pair pumped volume and time used will characterize the pipeline state.The hydraulic gradient has to exceed the maximum ground elevation transversed by the pipeline. As the hydraulic gradient decreases with the flow, there is a maximum flow which is patible with the hydraulic gradient exceeding the ground elevation at all pipeline point. Usually the critical point (minimum difference between the elevation and the hydraulic gradient)corresponds with peaks of the ground profile at the end of the pipeline.The decision variable is the dosage injected to each batch. Every node has six successors (0, 5, 10, 15, 20 and 25 ppm.). A node represents the state of the pipeline when a new batch is going to enter into the pipeline and the child of that node is the state of the pipeline when this batch is pleted pumped with a determined additive concentration. Notice that all the successors of a node have the same pipeline state but different DRA consumption and pumping time.The algorithm expands the most promising node at any time, that is the node with the minimum value of an objective function. The algorithm requires the extensive putation of states but the plex hydraulic equations of the simulator are too time consuming for be used during the search process. The presented algorithm performs several simulations with the simulator to obtain pipeline state data pleted with approximate hydraulic equation in the search algorithm.Previous simulationsA simulation of the plete batch sequence is puted at first considering no additive injection (0 ppm) in any batch. If the total simulation time is smaller than target time, then it is no necessary to use additive. An algorithm to set optimally pumps and valves can be used.A second simulation is executed adding 25 ppm to all batches. If simulation time is bigger than target time, then it is not possible to verify time constraint. Notice that the simulation is performed at maximum flow.At any other situation, the search algorithm is needed. A plete simulation is executed for each one of the discrete values of the DRA injection, considering that all batches are injected with the same quantity. For each simulation some data are stored in order to be used in the evaluation function and the approximate hydraulic equations.These data are the flow at the head, the batches situation along the pipe when a new batch is going to be pumped and the time when that situation has been reached. Notice that the batch situation is the same for the different simulations but the time is different because different DRA dosage has been used.文章出處: Dpto. Ingenieria de Sistemas y Automatica Universidad de Sevilla Seville, Spain減阻劑在輸油管道中的最佳使用摘要：管道公司需要空氣阻力減速添加劑（DRA）增加飽和管道的運(yùn)輸能力。在保證基本的產(chǎn)品規(guī)格的條件下，添加很低濃度的減阻劑后（百萬(wàn)分之5到25），管道的流量顯著增加。減阻劑的費(fèi)用太高，優(yōu)化它的耗費(fèi)是必要的。本文介紹了一種優(yōu)化算法，以盡量減少減阻劑的消費(fèi)，但滿足運(yùn)輸需要。該算法基于bestfirst搜索算法。本文中的一些結(jié)果是在西班牙北部的一條輸油管線中得到的。關(guān)鍵詞：運(yùn)輸系統(tǒng) 輸油管線 決策 調(diào)度 優(yōu)化 bestfirst搜索介紹石油產(chǎn)品從輸入節(jié)點(diǎn)輸入管道，這些輸入端通常是煉油廠或港口，輸出端為油庫(kù)。石油產(chǎn)品從儲(chǔ)罐中提取出來(lái)然后輸送給最終消費(fèi)者，輸送方式通常是汽車輸送。這些產(chǎn)品（柴油，汽油。）被注入到不同的批次，也就是說(shuō)，在任何時(shí)候不同的產(chǎn)品在輸送過(guò)程中都被相應(yīng)的接口分離開。輸送所需的能量由一組電泵提供。它已經(jīng)表明，摩擦阻力壓降或拖累，限制石油管道的吞吐量，可大大減少注射長(zhǎng)鏈聚合物， 所謂流動(dòng)改良劑或拖曳減速。. Toms的英國(guó)研究人員首次發(fā)現(xiàn)湍流減阻 。盡管在過(guò)去的40年里對(duì)減阻劑領(lǐng)域進(jìn)行了大規(guī)模的探查，但都沒(méi)有被普遍接受能夠解釋減阻機(jī)理的模型。一個(gè)很重要的關(guān)于高分子減阻劑的事實(shí)是它僅僅在湍流區(qū)起作用。聚合物改變了湍流的流動(dòng)特性通過(guò)降低湍流強(qiáng)度。事實(shí)表明，減阻劑系統(tǒng)在管子的核心只顯示了較少的波動(dòng)。然而，減阻劑不直接影響波動(dòng)核心，而是干擾突發(fā)過(guò)程并阻止波動(dòng)的形成。減阻劑的實(shí)際效益依很重要的。在同樣的壓降下，流體的流速更快，或是管線的壓降被降低后管道內(nèi)的流體仍然能保持原來(lái)的流速。流動(dòng)改良劑在1979年首次用于商業(yè)管線運(yùn)作中。凝膠型流動(dòng)改良劑曾被不同國(guó)家的石油運(yùn)輸公司所采用。這種類型的流體改良劑能達(dá)到30%的增幅。新一代的液體聚合物改良劑在保證產(chǎn)品基本規(guī)格的前提下能夠使流量增大100%。當(dāng)摩阻降低后，石油通過(guò)管道需要的能量減少。因此，降低能耗的同時(shí)能維持吞吐量，或是在不增加操作壓力時(shí)能夠提高吞吐量。減阻劑是一種昂貴的產(chǎn)品，所以它的使用只限于為了滿足用戶的需要而不得不提高輸送能力的工況下。然而，對(duì)于使用減阻劑的管道公司來(lái)說(shuō)盡量減少產(chǎn)品的用量是個(gè)很重要的問(wèn)題。管道的優(yōu)化問(wèn)題涉及到單流體輸送系統(tǒng)文獻(xiàn)中，例如水分配系統(tǒng)。其他的文章中描述了多流體管道的優(yōu)化算法，但減阻劑不能用在這些算法中。本文介紹了一種只有一個(gè)入口節(jié)點(diǎn)和出口節(jié)點(diǎn)（直接泵送）管線的算法以確定減阻劑最小用量及分配批次。此算法基于bestfirst搜索算法并借助于調(diào)度管道模擬器和近似算法得到。文章的安排如下：第2部分描述問(wèn)題并使用工具解決問(wèn)題。第3部分介紹優(yōu)化算法。第4部分提交了一條真實(shí)的管道的一些結(jié)果證明該算法的可行性。最后給出一些結(jié)論。優(yōu)化算法此算法基于bestfirst搜索算法?；旧?，該方法提供了一個(gè)搜索樹，搜索樹的節(jié)點(diǎn)描繪了系統(tǒng)（管線）的狀態(tài)，根節(jié)點(diǎn)是初始狀態(tài)，一個(gè)節(jié)點(diǎn)的接替點(diǎn)則是達(dá)到了一個(gè)決定行動(dòng)被應(yīng)用的狀態(tài)。獲得接替點(diǎn)的節(jié)點(diǎn)被稱作節(jié)點(diǎn)擴(kuò)展。我們的目的是通過(guò)搜索樹達(dá)到一個(gè)連接根節(jié)點(diǎn)的目標(biāo)節(jié)點(diǎn)。這些算法的主要問(wèn)題是當(dāng)節(jié)點(diǎn)擴(kuò)展的時(shí)候會(huì)組合激增，也就是說(shuō)，節(jié)點(diǎn)的數(shù)量或是計(jì)算時(shí)間會(huì)曾指數(shù)增長(zhǎng)。Bestsearch算法試著用賦值函數(shù)啟發(fā)式減少擴(kuò)展節(jié)點(diǎn)的數(shù)量。以這樣一種方式，在任何時(shí)候會(huì)擴(kuò)張的節(jié)點(diǎn)擁有最小的賦值函數(shù)。在這個(gè)算法的應(yīng)用上讓我們考慮主要的概念：管道狀態(tài)管道的狀態(tài)通過(guò)管道內(nèi)的不同位置描繪出來(lái)。如果給出切入點(diǎn)的泵輸量，就能確定管線的狀態(tài)。如上所述，在使用減阻劑的時(shí)候我們考慮最大流量，而且，在用泵輸送的時(shí)候所有可用的泵要是連通的。這樣，指定的流量是唯一的?？紤]系統(tǒng)的動(dòng)態(tài)行為，第二個(gè)關(guān)鍵的變量是時(shí)間。接著，兩臺(tái)泵的輸量和時(shí)間能描繪管道的狀態(tài)。管線的水力坡度必須超過(guò)最大的橫向地面高程。當(dāng)水力坡度隨著流量減小時(shí),將會(huì)有一個(gè)符合在管道的任何位置都超過(guò)地面高程的水力坡度。通常情況下，在管線末端，臨界點(diǎn)（海拔和水力坡度之間的最小差異）符合地形的最高點(diǎn)。每一批油品摻入的減阻劑的量是決定量。每一個(gè)節(jié)點(diǎn)有6個(gè)繼承者（百萬(wàn)分之0，）.每個(gè)節(jié)點(diǎn)代表管到的狀態(tài)，此時(shí)新的一批油品將要進(jìn)入管道，當(dāng)這一批油品摻入確定濃度的減阻劑并被泵送完后該節(jié)點(diǎn)的子節(jié)點(diǎn)代表此時(shí)管道的狀態(tài)。值得注意的是所有節(jié)點(diǎn)的繼承者都有相同的管道狀態(tài)但是減阻劑的耗費(fèi)會(huì)不同，泵送的時(shí)間也不同。該算法在任何時(shí)候都擴(kuò)展了最有前途的節(jié)點(diǎn)，也就是說(shuō)，該節(jié)點(diǎn)的目標(biāo)函數(shù)有最小值。該算法需要大量的狀態(tài)計(jì)算，但是模擬器的復(fù)雜液壓方程在搜索過(guò)程中耗費(fèi)大量的時(shí)間。 目前的算法運(yùn)用模擬器執(zhí)行一些模擬并配合近似液壓方程獲得管道的狀態(tài)數(shù)據(jù)。提前模擬一次完整的批處理模擬的順序是在任何批次中先計(jì)算不添加減阻劑（百萬(wàn)分之0）的情況。如果總的模擬的時(shí)間比目標(biāo)時(shí)間短，那么沒(méi)必要使用減阻劑。算法的作用是確定泵和閥門的最佳使用。接下來(lái)的模擬是往所有的批次中加入百萬(wàn)分之25的減