5 GEOMODELLING TRAPS TO AVOID

Joseph

作者：Neil Price

原文链接：https://www.linkedin.com/pulse/5 ... s-avoid-neil-price/


Over the past 20 years I’ve fallen into every one of these traps. Here’s what NOT to do.

1. RUSH IN. I get it, you want to add value. You may feel pressure from colleagues and management to show value too, given the investment in you and the software. However, jumping in and assuming the project goal will become apparent as you progress will lead to compromise, re-work, and diminish your role within the team. You’ll be reactive and start chasing deadlines, instead of proactively leading the project. Always remember, the objective isn’t to twirl a colourful map on your screen, it’s to build a model that helps inform decisions for your organisation.

2. INCLUDE EVERY DETAIL. One of the most common pitfalls and it usually accompanies #1. With no clear goal, people tend to build models at the scale of the data available (ie well logs). Their thought process is to cover their bases for any possible questions. Such models are unwieldy, difficult to update and costly in terms of time and expense. It's far better to work at a resolution tailored to the questions and problems you’ve been posed.

3. WORK IN ISOLATION. One of the really powerful benefits of reservoir modelling is sharing data and ideas interactively in and between teams. Models require buy-in across; philosophies, concepts, workflows, results and decisions. Don’t share these intermittently, you’ll risk swamping peers with too much information and alienating team members from making shared decisions. There are times to work and times to share. An effective modeller will give equal weight to both.

4. TRUST THE SOFTWARE. Effective modelling requires understanding of geological processes and geostatistics. Unfortunately, many geostatistical modelling methods are taught by rote. This leads to button pushing modelling where resultant parameters are merely ticked off a list. It’s far more powerful to build a simple model built on sound geological and geostatistical principles that you can defend, rather than a series of black-box outputs. Take the opportunity to learn some geostats theory and avoid becoming a button pusher.

5. ONE MODEL. One model to rule them all. Call it what you will; Reference, Base, Best Case. For many reasons the creation, delivery and assumption that one model represents the uncertainties, risks, and volumes of the reservoir has become ingrained in our modelling mindset. At a push there’s sometimes a high / low case thrown into the mix. All models are wrong, so why put all your eggs in one basket. A model is merely a representation of the reservoir’s possible characteristics and behaviour. Assess the uncertainty space early and build conceptual or ensemble models. Do not begin your modelling down one path and end in a cul-de-sac.

In conclusion, a good geomodel provides insight. It needs to add value by helping people make decisions. A good geomodeller:

• Clearly defines objectives
• Builds at the appropriate scale for the questions being asked
• Communicates regularly with their team
• Understands geological processes and geostatistical theory
• Frames the uncertainty space
  

#geomodelling #uncertainty #decisionanalysis #oilandgas

Joseph

【粗译】地质建模需要规避的五个陷阱


作者 Neil Price

经过20余年浸淫其中，我遭遇过这些类型的陷阱中的每一个，你绝不要重犯：

1、急切进入。我理解，你希望有所表现。你也许是从管理和同事那里感到压力，也想要展现你和软件的投入都有回报。不管怎样，激进和假定项目目标都显然会让你工作进程走向妥协、往复和削弱你在团队内角色价值。你将更被动和被最终检查点驱动，而不能积极引领项目。永远要记住，你的目标不是在屏幕上展示酷炫图案，而是要给你的公司、单位提供一个可资决策的模型。


2、极尽细节。最常见的陷阱之一，也是跟着第一条一同发生的。因为没有明确目标，人们倾向于构建一个尽可能细节化的模型（即，达到测井曲线精度的）。他们的想法过程是要把他们可能遇到的各种问题都涵盖在内。这种模型自然是蠢笨而难以更新，只是在时间和成本上增加消耗。更好的工作方式是针对你遇到的问题去定制一个解决方案（精度范围）。




3、无从协作。油藏模型的一个真正极其有用的地方就是可以交互性的在团队内和团队之间形成数据和信息共享。模型需要在专业互补、整体脉络、概念延展、工作程序、成果和决策等等方面进行妥协。不必阶段性的进行这些信息的共享，这会导致团队内和团队间疲于大量信息交换并且阻隔了团队成员协作制定决策。必须确保足够时间建模同时足够的时间分享，高效率的建模工程师懂得如何在两者间平衡。



4、迷信软件。高效建模需要的是对于地质过程和地质统计学的理解。很不幸，很多培训都是刻板教人学习地质统计学建模算法的。这导致了一种“跑流程式建模”，其中的参数只是在一个列表内抄过来。这还不如在你能理解的地质框架和地质统计学条件下构建一个简单模型，至少也不是一个黑箱输出。参加一些地质统计学理论培训并拒绝变成“跑流程建模者”。



5、包罗万象。一个模型包打天下。你也可以说这玩意儿：参考、基础、最佳。从很多原因上来说，创建、提供和假设一个模型可以表达不确定性、风险，油藏储量就会被我们因成见而固死。这急切需要在固化版本上增加一个高预期、低预期版本。所有的模型都是错的，所以为什么你要把所有鸡蛋放进一个篮子呢。模型可能只是油藏可能特征和状态的一个表达。在规划阶段接受不确定性空间并建一个概念化或是简化模型。不要在一条路上开始构建你的模型，并且一直跑到死胡同的尽头。



小结一下，一个好地质模型应该体现（对于油藏的）深刻理解。这种深刻理解是在帮助人们制定决策中体现价值。一个好的建模者：

• 清晰规划目标
• 根据问题制定合理模型精度
• 与团队紧密沟通
• 理解地质过程和地质统计学原理
• 预留处足够的不确定性空间
  

[color=rgba(0, 0, 0, 0.75)]#geomodelling #uncertainty #decisionanalysis #oilandgas