We are looking forward to this conference! It will be valuable for operators to attend and get ideas on how to leverage technology to find margin in the current and future environments. Look for the @revonosbiz booth to see how we use #ORMS to guide decision-makers toward making optimal decisions.
Operations Research (OR) is not new. The development of the field took off following World War II. Industry took up the pursuit not long after. Today, the Franz Edelman Award finalists for 2018 were announced. This award presented by INFORMS is named to honor Franz Edelman who was a pioneer of OR/MS. He led the Operations Research department at RCA Corporation, where he envisioned using mathematical optimization to manage business better.
A petroleum entity, China National Petroleum Corporation, is included in the finalists this year. Previous years have had petroleum companies make the list of finalists, like Chevron in 2013. Last year’s inclusion of BHP Billiton is the finalist that I know of nearest to E&P operations, but it is in mining production, not the petroleum business. The common thread in the petroleum sector finalists has been midstream applications. CNPC applied ORMS to natural gas pipeline networks. Chevron optimized refinery operations.
Optimization methods are applied to E&P operations by companies such as ExxonMobil and BP. I applied OR to hauling at Noble Energy and discussed the models in my thesis. I believe this is rare. The reason I propose has to do with the common dialog that accompanies the historically volatile oil & gas markets. Producers are too happy to keep business as usual when prices are high. The most common response to low prices is to renegotiate contracts, drill less or not at all, pinch pennies and await the “recovery”. The shale explosion has been fueled by technological improvements that make development economic at lower prices. OR can help to make production operations economic at lower prices. A nice side-effect to making production operations more efficient is lowering the energy input to get the production output, reducing pollution.
I want to see OR topics become common in the conference rooms of E&P companies of all sizes, not only the majors. Let’s apply mathematical optimization to production operations and reduce LOE to make more profit.
Description of Operations Research
Operations research (OR), operational research, management science and decision science are just some of the names used for this discipline that seeks to arrive at answers to complex decisions faced by organizations of all sizes. OR is characterized by seeking efficiency in allocation of scarce resources. This means that OR applies to everything.
When trying to solve problems or make decisions, OR provides an approach that can be applied to deliver results. This is in contrast to the definition of Data Science which primarily focuses on gaining insight from data. Individuals can choose to make decisions based on this insight, but the process of improvement is not a definitive part of data science. OR incorporates the process of improvement and is commonly known as the “science of better”.
Operations Research Methodology in (Data) Science Terms
The Operations Research Methodology approach to decision-making is familiar to the data scientist. The methodology is an application of the scientific method. The OR professional is presented with a problem, gathers data, formulates hypotheses, tests the hypotheses and generates useful solutions. There are three major disciplines or processes within the Operations Research framework.
Data Collection, Gathering and Exploration is an integral process in OR. An OR professional may not be given any raw data to begin with, but must seek data for input to define the problem mathematically.
Exploration of the data overlaps into the second discipline within OR. This is the Data Science portion of OR. Data Science’s main purpose is to gain additional insight into a process. Data Science is the practice of finding insights in the data.
In order to find insights in the data, the data scientist must formulate a hypothesis utilizing the data at hand. Manual or semi-automated modeling is used to formulate hypotheses. Combining the raw data and additional insights from the data exploration step, mathematical models can be built for simulation, optimization or other decision support purposes. These models become the hypotheses to be tested.
Hypothesis testing is regular practice in the data scientist’s day-to-day work. For effective hypothesis testing, the OR professional should utilize a mathematical model that is a simplification of the actual, real-world process. A test of these models can be functional tests, like validation used in regression or classification models, or alternately running simulations or implementing real-world tests.
The final key portion of OR is Decision and Outcome Management. The field of Decision Management offers methods and systems to track of decision points and the choices made. It is important to provide actionable information through data science, but it is equally important to track the recommendations or decisions available as well as the actual decision that is made, and the effectiveness and results of the decision outcome. In this way, the results can be reincorporated into data collection, and subsequently model improvements.
Operations Research is composed of multiple fields and disciplines. Effective Operations Research leverages the skills of these different fields to improve processes and outcomes. The fields and terms used in OR are maturing, and this is leading to greater definition and understanding of the various components of an effective OR outcome. For example, Data Science is a key component of OR, but does not capture all of the components of that methodology. Statistics and business analytics are a key components of Data Science, but non-statistical methods like machine learning are key tools in the Data Science discipline. In order to find the most efficiency in allocation of scarce resources, all components of Operations Research should be employed.
There has been a lot of “Big Data” talk in industry for several years now. If you work for any size corporation, you most likely have heard the words “Big Data” or seen a demo or observed the glazed over look on the faces of management when the term comes up. The last 5 years have been spent educating the majority of the market on the definition of Big Data. Additionally, there have been strong efforts to peel back the covers within companies and reveal the massive treasure trove of data sitting on servers, desktops, laptops, and other sources. Connecting the dots of how Big Data intersects the data within corporations is where a corporation’s competitive advantage will come from in the coming years. Those willing to make this connection a priority will be the companies that can do more with less, and gain key insights to make the next step change in their operations.
Big Data, at its core, describes a volume of data that is beyond traditional information technology’s capability to store, manage, analyze and process efficiently. Most companies don’t have, or are just investing in, new technology to handle the data they have on premises, so most companies are in a situation where they already have big data. Other definitions mention high volume/velocity/variety data. Again, with the many automated systems, equipment, applications, and 3rd party data feeds, companies will find that they’ve entered the big data landscape, whether they realize it or not.
The market for Big Data is expected to exceed $60B US by 2020 (from $7.6B in 2011). Executive teams are getting pressure to do something with Big Data, and several companies are chasing technology – putting in big data platforms for millions of dollars, but not understanding its use or how to best gain value from their data assets. Several vendors are entering the space both for application solutions, but also for consulting expertise to help companies navigate this new technology option.
The Big Value from Big Data is the ability to effectively combine Information Technology (IT) with Operational Technology (OT). In addition, this technology is enabling companies to combine and democratize the data in their siloed disparate best in breed operations applications with one another to reveal new insights in the hands of the user base within a company. The Big Data platform allows these advances to happen, and also has the ability to move companies towards new levels of analytics – from descriptive, to predictive and on to prescriptive.
So, why should you care?
- Companies have continually improved operations through modernization of equipment, adopting new technology for data capture and reporting, automating data feeds, and applying process improvement methodologies to cut waste and add value. The effect of improved operations can be seen in increased productivity and safety, decreased energy usage and cost, better situational awareness and better two-way insight with customers, partners and vendors. Big Data and the opportunities it enables will be the next step change in improvement to operations.
- Competitors are going to embrace the technology, particularly as it gets more approachable. Companies that had good data strategy and stewardship foundation will gain value sooner, but many companies are on a relatively level playing field when it comes to stewardship (poor), so developing a data strategy and reviewing your own data asset health now is key to quick wins and potentially gaining a competitive advantage in the near future.
- Technology is always improving, and the adoption of paradigm shifting toolsets is inevitable.
- The change involved in adopting Big Data and Advanced Analytics within your organization will involve culture, procedural and technologic shifts in the way your company operates. Developing a plan now will ease some of the growing pains that come with this new technology.
There are several compelling reasons (and value cases) to adopt this new technology. Big Data and its value doesn’t have to be scary. In fact, developing a plan that focuses on your operations priorities will make the transition relatively painless. Big Data complements your existing applications and systems, and makes them more valuable and more powerful in the hands of your employees. Don’t ignore the value that you’ve accumulated in your Data Assets over the years. Unleash it!
The different levels of analytics can be summarized into three categories. These are descriptive, predictive and prescriptive analytics.
The first category, descriptive analytics, is necessary to be able to perform analysis. The tools and processes that belong in this category are not necessarily simple, but are sometimes only simple statistics describing data. Traditional business intelligence (BI) falls under this category. The products of data exploration during data science does as well. Interestingly enough, even “alarming” falls under this category. Many times, alarms are considered advanced tools. They are certainly valuable, but really only tell a business what has already happened based on hard data and rules. The reason analytics in this category are necessary for being able to perform analytics in either of the other two categories is that achieving descriptive analytics prepares the organization’s data for consumption. Even if analytics implemented in systems are not part of an organizations regular business, personnel take in the descriptive analytic products and perform additional processing themselves to determine decisions. Another area that falls under descriptive analytics is data exploration. Data exploration is a manual, iterative process that generates statistics, which lead to insight on how to build models useful for implementation in a process. This is the data scientist’s first step.
Predictive analytics are the more common of proactive analytics. This category is vast. In statistics, prediction is also known as estimation and not necessarily forecasting. Predictive analytics are concerned with unknown quantities and factors. Factors can be flags, levels, classification or clustering. Predictive analytics are performed with any of a multitude of tools. A simple example is ordinary least squares (OLS) estimation. The machine learning wave mostly lives in this category. Hierarchical & k-means clustering, random forest regression & classification, support vector machines and ridge regression are some of the tools that exist in this category. These tools each have their strengths and weaknesses and it takes a learned person (or automated evaluation of each one) to figure out which is the best to apply to a specific problem. These problems include finding production excursions, detecting fraud, estimating power output, demand forecasting, risk estimation and claims prediction. This list is barely a scratch on the surface of the use cases for predictive analytics.
The final category, prescriptive analytics, is the other proactive category of analytics shared by two general classes of tools. Machine learning is one of these. Operations research is the other. Each of these classes makes extensive use of mathematical optimization techniques. Artificial neural networks and deep learning methods are machine learning techniques which are capable of recommending decisions in certain applications. The problem domain for deep learning is continually expanding. This technology is what makes autonomous vehicles possible. There are other techniques that can be applied to recommend decisions through mathematical optimization, such as stochastic gradient descent and other unconstrained minimization techniques. When the rules of the decisions are known ahead of time, a machine learning method can be applied to recommend a decision. Many times, decisions must be made within more general constraints instead of rules. For instance, a supplier might need to deliver product from different plants to a variety of depots. This scenario has demands for product at each depot and maximum supply at each plant. The transportation costs between plants and depots can be taken into account and a minimum-cost solution found using linear programming. This use case is one that has had an efficient mathematical solution since World War II. The applications of linear programming have been explored and expanded throughout many industries. It can be applied to manufacturing (blending, for instance) and finance (using portfolio balancing) for two examples.
These categories of analytics can be regarded as a progression, but they need not be, with the exception of having a handle on descriptive analytics before beginning any proactive analytics. The two proactive categories can overlap some as far as techniques, but one distinguishing characteristic is the level of additional insight that must be obtained to make a decision. For instance, if a rule is applied to results of a machine learning method, then the method is not prescriptive. Whereas, if the results of a technique are decision points by themselves, then it is prescriptive.