Reservoir engineering: tasks across the life of a field

Oil and gas lie thousands of metres down, in the pores of rock. We cannot see the whole reservoir. We judge it from a few wells, from seismic sections (pictures of the layers made with sound waves) and from the way the reservoir responds to production. Between the wells lie hundreds of metres that nobody has seen. The reservoir engineer's task is to build, from scattered measurements, a picture on which expensive decisions can rest.

The work comes down to three questions and one constant duty:

1. How much? — the hydrocarbons in place: the oil initially in place and the gas in the reservoir. We calculate them from the rock volume, the porosity (the share of the rock that is pore space), the saturation (the share of the pore space filled with oil, gas or water) and the fluid properties (Volumetric estimation of hydrocarbons in place).
2. What part? — the recovery factor (the fraction of the oil in place that we can produce) and the recoverable reserves. The fraction depends on the rocks and fluids, on the natural energy of the accumulation and on how we develop it (Reservoir drive mechanisms, Why water is injected: pressure maintenance and oil displacement).
3. How fast? — the production profile: how much oil, gas and water we will produce year by year, how many wells that needs and how large the facilities must be (Well inflow: the well performance diagram, the profile in the project model).
4. And throughout, reservoir management: comparing actual performance with the forecast, explaining the differences, and making proposals for the wells and for how they are operated.

No single formula gives the answers. They rest on four groups of knowledge: about rocks, fluids, flow and reservoir energy. On top of these come calculations of displacement (how one fluid pushes another out of the rock) and simulation (a computer model of the reservoir). In the library each group has its own lessons.

A field goes from the first exploration well to the abandonment (closing and sealing) of the last well. It is convenient to divide this path into four stages. Each stage has its own main question, its own data and its own decisions. The boundaries between the stages are a matter of convenience. We estimate the reserves throughout the life of the field, and we start to collect data for the project already during exploration.

A single reservoir shows how such work looks in numbers. The same illustrative oil reservoir runs through the lessons of this section. Its pore volume (the volume of all the pores) is 16.0 million m³ and its initial water saturation is 0.25. Its initial pressure is 30 MPa. Its bubble point pressure is 18 MPa at 85 °C. Different lessons solve different problems for it, and the results agree with one another:

After a discovery the reservoir engineer has hardly any data of their own: one or two wells, seismic maps, well logs, a few metres of core, fluid samples and a well test. From these data the engineer must answer one question: is it worth spending more money?

• Oil in place and its range. We take the rock volume from maps (Gross rock volume from maps) and the porosity and saturation from well logs and core (Log interpretation workflow, Core analysis). We calculate the oil in place by the volumetric method (rock volume times porosity times oil saturation, converted to surface conditions) and give it at once as a range (Volumetric estimation of hydrocarbons in place, Probabilistic volumetrics by Monte Carlo simulation).
• Fluids. Samples give the fluid type, the bubble point pressure, the solution gas–oil ratio and the formation volume factor (Reservoir fluid sampling, Laboratory PVT tests). An error in the formation volume factor carries over in full into the oil in place.
• Pressures and contacts. A wireline formation tester (a tool lowered on a cable to measure the pressure in the rock at chosen depths) gives pressure gradients, the positions of the contacts (the surfaces between oil, gas and water) and the first sign of whether the layers are connected (Pressure measurement with the wireline formation tester, Pressure gradients, fluid contacts and capillary equilibrium).
• Productivity. A well test is a short period of production followed by a shut-in of the well. It shows what rate the reservoir can deliver. It also lets us estimate the permeability, the skin factor (a number that shows how much the zone right around the well helps or hinders flow) and sometimes the distance to the boundaries (Well testing: purposes and types of test, Pressure build-up analysis).
• Analogues. What we have not measured, we take from analogue fields. These are fields with similar rocks, fluids and reservoir conditions. This is only a first estimate, and we still have to check it.

The main result of this stage is not a single figure for the oil in place. It is a range, together with an understanding of which uncertainty affects the decision most. Appraisal wells (wells drilled to find out how big the discovery is), 3D seismic (seismic data that give a three-dimensional picture) and extended well tests (longer trial production) cost money and time. They are worth doing only if they can change the decision. We decide which data to buy, and when, with a decision tree and the value of information. How appraisal work narrows the range is covered in the lesson on technical uncertainty.

Once we have decided that the accumulation is worth developing, the reservoir engineer and other specialists look for the best way to do it. A development option is a set of decisions:

• Source of energy. Rely on the natural energy of the reservoir, or inject water or gas from the first years (Reservoir drive mechanisms, Why water is injected: pressure maintenance and oil displacement, Displacement of oil by gas).
• Wells. How many wells, where and of what type — vertical, deviated (slanted), horizontal, hydraulically fractured (the rock around the well is cracked open with high-pressure fluid) (Horizontal wells, Hydraulic fracturing: purpose and well selection) — and in what order to drill them.
• Rate of withdrawal. The height and length of the production plateau, and the capacity of the facilities that will deliver it (production profile).
• Phasing. We can build everything at once, or start with the better-understood part of the accumulation and decide on the rest from the data of the first wells.

For each option we build a forecast of oil, gas and water production and of injection volumes. At early stages we use simple calculations: material balance (General material balance equation for oil reservoirs), a forecast for the drive mechanism of the reservoir (Performance prediction for solution gas drive reservoirs) and a displacement calculation (Buckley–Leverett theory and the Welge method). For heterogeneous reservoirs (reservoirs whose properties vary from place to place) and complex systems we use a reservoir simulation model (What a reservoir simulation model is and why it is needed). We compare the options on their economics — net present value (the value today of all future income less costs) — and on their risks (Sensitivity analysis and scenarios).

Uncertainty at this stage is still large, and a good project deals with it openly. It is useful to ask not what we do not know about the reservoir, but which gap in our knowledge could change the decision:

How to arrange the surface facilities of the field is a separate topic (Field development concepts: surface facilities). It covers where to put the gathering points, how to treat and transport oil and gas, and what to build onshore and offshore.

When production starts, the reservoir for the first time responds to being produced on its full scale, and now the most informative data arrive. We call the systematic collection and analysis of these data reservoir surveillance:

• oil, gas and water rates of every well, the gas–oil ratio and the water cut
• reservoir pressure, found when wells are shut in (Pressure build-up analysis, Build-up in a bounded reservoir and average reservoir pressure), and layer-by-layer pressure measurements in new wells, which show which layers are being drained (emptied of oil) and which are not
• inflow and injection profiles along the wellbore (how much each interval produces or takes), and the positions of the contacts, from production logging (measurements made in working wells)
• composition and properties of the production, injection volumes and pressures

We compare these data with the forecast. A complete match is rare, and the engineer must understand why. Is the reservoir worse or better than we thought? Did water arrive along a thin layer? Did a fault divide the accumulation into blocks? Are the wells working with a damaged near-wellbore zone (the rock around the well, which drilling or production can clog)? The answer changes both the model and the plan. We drill additional wells, move wells to other layers, change injection, and carry out well treatments such as matrix acidising and hydraulic fracturing (Formation damage and damage skin, Acid treatments: principle, economics and candidate selection, Hydraulic fracturing: purpose and well selection).

New data also improve the description of the reservoir. Material balance on the pressure and production history checks the oil in place and whether water is flowing in from the aquifer (water influx) (General material balance equation for oil reservoirs, History matching of aquifer parameters). We adjust the reservoir simulation model until it reproduces the history (History matching). A good match is necessary for a reliable forecast, but it is not enough: the same history often fits different models.

The simplest way to extend a forecast is to continue the production trend along a decline curve. This is quick and needs no model. But the curve knows nothing about the causes of the decline, and two or three points do not fix its shape. More is in the advanced section on decline curves.

Sooner or later the production of every accumulation begins to fall. The pressure declines, water or gas reaches the wells, and their share in the production grows. We slow the decline with artificial lift (pumps and other equipment that help to lift the liquid up the well) (Artificial lift: when it is needed and how to choose the method), with new wells between the old ones, by redistributing injection, and by treating the near-wellbore zone. According to how we act on the reservoir, we usually divide production into three stages of recovery:

Primary recovery

Production using only the natural energy of the reservoir: compressibility, solution gas, the gas cap and the aquifer (Reservoir drive mechanisms). Artificial lift belongs here too. A pump helps to lift the liquid up the well, but it adds no energy to the reservoir.

Secondary recovery

Injection of water or gas to maintain the pressure and displace the oil (Why water is injected: pressure maintenance and oil displacement, Displacement of oil by gas).

Tertiary recovery — enhanced oil recovery methods

Enhanced oil recovery (EOR): methods that change the properties of the fluids or the way they interact with the rock. They are thermal (steam, in-situ combustion, which burns part of the oil in the reservoir), miscible gas (carbon dioxide or hydrocarbon gas that mixes with the oil), chemical (polymers that thicken the water, surfactants that lower the surface tension) and microbial. A broader idea is improved oil recovery (IOR): increasing recovery by any means, including new wells and injection management.

The stages do not have to come in order. We often start waterflooding in the first years of development, and we develop accumulations of high-viscosity oil (very thick oil) with steam from the very beginning.

To choose how to continue, we need to know where the oil is left and in what form. After waterflooding it remains in two forms. Residual oil consists of droplets that capillary forces (the forces of surface tension in narrow pores) have trapped in the swept pores. Only methods that change those forces can recover it. Bypassed oil is the oil in the parts of the reservoir that the water has not reached (in Russian-language practice, unswept “pillars” of oil). It is mobile (it can still flow), and we can reach it with new wells or by redistributing injection.

When production no longer pays for its own costs, we reach the economic limit (how it is calculated). Then the wells are abandoned (closed and sealed), the facilities are taken down and the land is reclaimed (restored for other use). The cost of abandonment is an obligation, and we build it into the economics of the project from the very beginning (abandonment in the project model).

No two reservoirs are the same. Accumulations in the same horizon (the same rock layer) on neighbouring uplifts (raised parts of the structure) can differ in the structure of the trap (the rock structure that holds the oil), in layering, in oil properties and in aquifer strength. They can also respond differently to the same development system. So a project copied from a neighbouring field without changes is a frequent source of errors. The experience of analogues is a starting point, which we check against the field's own data.

Here, incomplete knowledge is not an oversight. It is a condition of the work. Every new well tests the map at one point, and such tests often change the map. They shift a contact, or find a fault that divides the accumulation into blocks with different pressures, or find a layer along which the injected water will escape. We cannot wait until the description is complete, because the money we invest does not come back until there is production. So we make decisions on incomplete data — but in such a way that we can correct them when more data arrive.

• We keep the estimate as a range and revise it with every new set of data (Geological and technical uncertainty of a project).
• We make decisions that are hard to change later, on more data, and easily corrected ones earlier. Development begins with the better-understood part of the accumulation.
• We compare a reservoir with analogues by physics (rocks, fluids, drive mechanism, development system), not by its position next to them on the map.
• The project includes data gathering: pressure measurements, well tests, coring and sampling in new wells. Data are cheapest to get while the well is being drilled.

The reservoir engineer gets almost all the input numbers from other people: geologists and geophysicists, petrophysicists, laboratories and well testers. The engineer's job is to combine these data into one consistent reservoir model, and to turn them into forecasts. Others will use the forecasts to drill, to build and to count the money. So it is important to understand where the numbers come from, how far we can trust them and what will be done with them next.

The links work both ways. The development forecast sets the number of wells and the capacity of the facilities. The limits of the facilities — for example, how much water a treatment plant can handle — in turn become limits on the forecast. A good forecast is agreed with everyone who will build and pay on its basis.

The common language of exchange is units and reference conditions (the temperature and pressure to which volumes are referred). We cannot add a forecast in barrels at 60 °F to measurements in cubic metres at 20 °C without converting. And we cannot compare a pressure read on a wellhead gauge with a pressure from the model without correcting for atmospheric pressure (Systems of units: SI, oilfield and metric oilfield).