微信一键登录

科吉思石油论坛

搜索
查看: 2454|回复: 0
打印 上一主题 下一主题

【FAQ】2.3 时间步的影响

[复制链接]

136

主题

205

帖子

264

积分

中级会员

Rank: 3Rank: 3

积分
264
跳转到指定楼层
楼主
发表于 2020-6-17 14:29:41 | 只看该作者 |只看大图 回帖奖励 |倒序浏览 |阅读模式
本帖最后由 张鹏326 于 2020-7-14 11:36 编辑

2.3 Influence of timestep
2.3 时间步的影响

Calculation and reporting steps
计算和报告时间步
报告步的产量
Often during one reporting step (where the well conditions are set) the simulator passes several calculation steps, the size of these steps is chosen to optimize approximation and convergence – picture 45.
通常,在一个报告步(设置了井条件)中,模拟器会通过几个计算步,选择这些步骤的大小可以优化逼近度和收敛性 – 图45。
The well rate on a reporting step is rate on step 3 (picture 45). As a result, well rate depends on time step sizes during reporting step.
报告步骤的井产量是三个(计算时间)步的产量(图45)。因此,井产量取决于报告步骤中的时间步长(即计算时间步)。

Current indices uncertainty
当前索引不确定性
The simulators report all current measurements (such as rates, water cut, BHP etc) for last calculation step during current reporting step, so the values depend on size and number of calculation steps during current reporting step. This is a function of:
模拟器会报告当前报告步骤中最后一个计算步骤的所有当前测量值(例如产量,含水率,BHP等),因此这些值取决于当前报告步骤中计算步的大小和数量。与以下内容有关:
  • Limits on Max and Min timesteps.
  • 最大和最小时间步的限制。
  • Settings of non-linear and linear problem solution – accuracy, iteration number limits, linear system algorithm choice.
  • 非线性和线性问题求解的设置 - 精度、迭代次数限制,线性系统算法的选择。
  • Number of cores used – since linear problem solution algorithm changes.
  • 使用的核数 – 由于线性问题求解算法发生了变化。
Instead of Current data, it is better to use either Cumulative, or Average for this step. For example, average rate for the step is cumulative production during this time step divided on its length.
(查看时)最好使用“累积”或“平均值”,而不是“当前”数据。例如,该步骤的平均产量是该时间步骤中的累积产量除以时间步长。

The current well modes are uncertain
目前的井控方式不确定
If the well has several limits, like, liquid production, BHP and THP, these limits are checked on each calculation time step. As a result, for different calculation settings, there will be different timesteps and different well switch times.
如果该井有多个限值,例如产液量、BHP和THP,则在每个计算时间步都会检查这些限值。因此,对于不同的计算设置,将会有不同的时间步长和不同的井开关时间。

Well closure time is uncertain
关井时间不确定
If the limits make the well shut in (like, BHP above block pressure), for different time steps the well may be shut in at different moments of time. In this case, the cumulative indices also change.
如果由于限制使井关闭(例如,BHP超过网格压力),对于不同的时间步骤,井可能在不同的时间被关闭。在这种情况下,累积指标也会发生变化。

Conditions checked on calculation time steps
计算时间步的检查条件
If some condition is checked on every time step, and if it is fulfilled, the well regimes change, then at moments depending on the following settings:
如果某些条件在每个时间步都会检查,并且满足条件,则井的状态会发生变化,取决于以下设置:
  • Min and Max timestep limits,
  • 最小和最大时间步长限制,
  • Non-linear and linear problem settings,
  • 非线性和线性问题设置,
  • Number of cores
  • 内核数量

The model behavior can change by a «jump» . This is how small rounding errors produce large differences in calculation.
模型行为可能通过“跳转”来改变(注:即很小的扰动)。 这就是小的舍入误差如何在计算中产生很大差异。

Example.
示例。Economic limits – WECON (see 13.20.76)
经济极限 - WECON(请参阅13.20.76)
Typical case: well water cut or Gas-oil ratio is checked and the well is closed (WECON, see 13.20.76). Here two uncertainties influence:
典型情况:检查井的含水率或气油比并关闭井(WECON,请参阅13.20.76)。这里有两个不确定性影响:
  • Current value of parameter is used;
  • 正在使用的参数当前值;
  • It's checked on calculation time step.
  • 在计算时间步上进行了检查。
As a result of WECON, global model changes – the well or perforations close.
WECON的结果是,整体模型发生了变化 – 井或射孔关闭。

Example.
示例。
On January 1 well has water cut 99.01% on 1 core and 98.99% on 8 cores. In the former case the well is closed, in the latter it may work for one more year (calculation step) and produce 100 m3 of oil, that is, 30% of its yearly production on the previous year.
在1月1日,一口井的含水率在用一个内核上(运算时)为99.01%,八个内核上为98.99%。在前一种情况下,井会关闭,在后一种情况下,井可能会再生产一年(计算步),并生产100立方米的石油,即上一年年产量的30%。
Which answer is «more correct»?
哪个答案“更正确”?

How do we cope with this uncertainty?
我们如何应对这种不确定性?
  • As with one of possible scenarios.
  • 就像多个可能情况之一。
How to reduce uncertainty?
如何减少不确定性?
  • Calculate several scenarios, with different maximal time step limits.
  • 计算具有不同最大时间步长限制的几种情况。
  • Use WTEST (see 13.20.186), designed to reduce uncertainty influence on global model (well closure). The WTEST (see 13.20.186) may reopen the well. Also, it's checked on calculation time steps so it's subject to uncertainty too.
  • 使用WTEST(请参阅13.20.186),其目的是减少不确定性对整体模型(关井)的影响。WTEST(请参阅13.20.186)可以重新打开井。此外,它还会检查计算时间步长,所以也受到不确定性的影响。

Compensation – WCONINJP (see 13.20.46)
补偿(注:注采平衡)– WCONINJP(请参阅13.20.46)
Injection well rate is calculated from production well rates. There are two uncertainties:
注入井注入量由生产井产量计算得出。有两个不确定性:
  • Current parameter value used;
  • 正在使用的参数当前值;
  • The check is on calculation timesteps.
  • 在计算时间步上检查。
The result is global model change, because injection well regime changes.
结果是整体模型发生变化,因为注入井状态发生了变化。

Group control – GCONPROD (see 13.20.88)
井组控制 – GCONPROD(请参阅13.20.88)
Group production is distributed between wells – members of group. There are several uncertainties:
井组产量在各个井组的成员(井)之间分配。有几个不确定因素:
  • current parameter value used;
  • 正在使用的参数当前值;
  • the check is on calculation timesteps;
  • 在计算时间步上检查。
  • wells in group can be calculated by different cores, then answer depends on order of summing.
  • 井组中的井可以由不同的核心计算,然后答案取决于求和顺序。
The result is global model change, because injection well regime changes.
结果是整体模型发生变化,因为注入井状态发生了变化。

Group control and others
井组控制及其他
  • Economic limits for groups;
  • 井组的经济极限
  • Surface networks;
  • 地面管网
  • Triggers (ACTIONX, see 13.20.164).
  • 触发器(ACTIONX,请参阅13.20.164)。
  • ACTIONX (see 13.20.164) can generate most unstable calculations since it can set:
  • ACTIONX(请参阅13.20.164)可能会生成最不稳定的计算,因为它可以设置:
  • Well mode changes,
  • 井控方式的改变
  • Group mode changes,
  • 井组控制方式的改变
  • Surface network regime changes,
  • 地面管网状态的改变
based on current value of parameter on one well. This global model change due to one current value of parameter on calculation time step is really a winner among all instability and uncertainty options.
基于一口井的参数当前值。在计算时间步上,由于参数的一个当前值而导致的全局模型更改,实际上是所有不稳定和不确定性选项中的获胜者。

(注:2.3本质上与2.2的内容没有区别,只是从时间步方面进行了解释。
总体而言,数值模拟的计算分为报告时间步和计算时间步,正常情况下在进行数值计算时,若是大的时间步没有满足收敛条件,则会将其拆分为多个小的计算时间步来细化计算。很多因素(单井控制、井组控制、ACTION系列、经济极限)都可能对计算时间步的分配有影响,而井控方式等生产制度会依据在计算时间步的计算结果进行判断、改变;如2.2种所描述的,内核数量也会影响计算结果,等等。在多个因素的综合确定下,数模计算结果有着一定的不确定性。
通常可以通过:设置内核数量、应用WTEST关键字、调整最大最小(计算)时间步限制等方式来减小这种不确定性。--qcy)

回复

使用道具 举报

您需要登录后才可以回帖 登录 | 立即注册

本版积分规则

Archiver|手机版|小黑屋|服务支持:DZ动力|科吉思石油技术咨询有限公司 ( 京ICP备15057753号

GMT+8, 2024-5-1 07:57 , Processed in 0.343750 second(s), 34 queries .

Powered by Discuz! X3.2 Licensed

© 2001-2013 Comsenz Inc.

快速回复 返回顶部 返回列表