Ryder Scott Company Petroleum Consultants September-November 1998, Vol. 1, No. 1, Story 3

Newsletter

Model saves millions of dollars in development capital

Ryder Scott, working with the Nuevo Energy Co. engineering staff, has been successful in optimizing well-spacing requirements from full-field modeling results for the Santa Clara Lower Repetto oil field.

"Development-capital savings can be measured in the millions of dollars as well as improved economics on the remaining development plan," said Tom McCollum, senior exploitation engineer at Nuevo.

A key element in the engineering study was the modeling work performed by the Ryder Scott simulation team headed by engineer Dean Rietz. The team developed a basic black-oil model of the Lower Repetto reservoir and concluded that a proposed 28-well infill-drilling program to recover 800,000 barrels per well would not optimize field development.

"By increasing initial spacing requirements, we were able to capture more reserves per well and do a better job in controlling the initial reservoir pressure depletion around each wellbore."

Based on initial model results, Nuevo and Ryder Scott drew up a recovery plan that included strategically locating about 12 wells in the 8,000-ft-deep field, which is located in 200 ft of water in the Santa Barbara channel seven miles offshore Ventura, California.

"By increasing initial spacing requirements, we were able to capture more reserves per well and do a better job in controlling the initial reservoir pressure depletion around each wellbore. That gave us overall better economics per remaining development well and, as an added bonus, a longer per-well life that increased overall recoveries from the reservoir by almost five-million barrels," said McCollum. "Usually when you drill less wells, you expect to get less overall reserves. However, in this case, the model showed us the opposite. Without the modeling results, Nuevo could have very well over developed this reservoir."

Considering each well costs $4.5 million to $5.5 million to drill and complete, implementation of the new recovery plan will save more than $50 million by eliminating overdrilling. Development will be accelerated because the plan does away with the 45 to 60 days per well it would have taken to drill and complete the unnecessary wells.

The reservoir model is also showing an uncanny accuracy for predicting production of the Lower Repetto sands. Nuevo initially tested the model by comparing predictions to the actual production of the Santa Clara S-62 well. Production came within 90 percent of the modeling prediction. The operator then drilled the S-28 well to the east and ran the simulation with the new well as an additional test before producing S-28. Again, the model prediction to actual production came within 90 percent.

At first, S-28 was modeled with simulated production from two of the lowest four sands, M and N. Initial production was only half of the potential production indicated by the model.

After sufficient pressure drawdown was experienced in the M zone, the lower most N sand cleaned up and production peaked at 760 B/D. At that point, the well began behaving as the model initially predicted. S-28 hit a peak at 804 B/D before stabilizing at 780 B/D and then declining. The model showed initial production at 700 B/D with peak production averaging 757 B/D with both sands open.

In late July, the well was making 508 B/D and the model was tracking closely to that figure, said McCollum. "You can’t get better estimates and our confidence in the model is growing each time we use it," he added.

Rietz said one of the reasons that the model behaves as the field does is the Ryder Scott team carefully per-formed a time-consuming history match and made sure that the pore volumes and the quantities and descriptions of the fluids in place were correct. From log data, the group initially determined the structure tops for each sand as well as porosity and permeability and the net-to-gross ratio of sand for each well.

"You can’t get better estimates. Our confidence in the model is growing each time we use it."

—Tom McCollum, Nuevo Energy Co.

The engineering work involved preparing PVT data, generating relative-permeability relationships and preparing fluid contacts and completion/production histories. Once those were assembled, the team calibrated the model through history matching.

The model has thus far proved to be very accurate, but its beauty also lies in its simplicity and lower cost to construct. Although the multilayer model contains variable thicknesses of the beds, Rietz assigned homogenous reservoir properties to each of the five layers, including a constant porosity and permeability. Typically, a homogenous reservoir model has optimistic recovery predictions, however, in this case, the model served as a very effective tool for analyzing field-development options.

Although Ryder Scott has the capabilities to incorporate detailed geological modeling, including geostatistics, the firm intentionally kept the model simple. "In this case, we felt that we could provide a simplified, full-field model that would provide the details necessary to complete the client's objectives," said Rietz. "Regardless of the size of the field or budget, the client';s objectives and the field characteristics dictate the complexity of the study."

Nuevo took advantage of the model's utility. "There is a skepticism about predicting future production from a model," said McCollum. "Usually you have to make several runs and struggle with different parameters before the results are reasonable. However, this time, we hit it on the head. No one in our office has seen anyone use the results of a model like our group has. Most simulations I have seen are done through a black box in a research department and then shelved."

Rietz was involved in the internal decisions of McCollum’s group, which outsourced the simulation work to Ryder Scott. "The key to this and other modeling successes is working closely with clients and giving them workable models," said Rietz.

Before doing the modeling work, McCollum and Rietz agreed that they should avoid overengineering it and should instead concentrate on such fundamentals as determining original oil in place, delineating geological structures and identifying water encroachment. "Many models are overanalyzed and too many variables are put in the model in an attempt to control it," said McCollum, "Then, if an overly complex model does not match up, it’s very difficult to find what needs to be corrected."

In late July, Nuevo was starting completion work on the S-61 well with anticipated first production in early September. The simulation for S-61 indicated the well should come in at about 600 BOPD.

"Ryder Scott is in the business of predicting and has a good track record, but it’s a difficult job to put yourself on the line every time with a prediction," said McCollum. Even the most carefully constructed computer models don’t always accurately predict real-world behavior.

"Luck is always a factor in how close any modeling predictions are to actual performance. In Nuevo’s case so far, we have been a little lucky. However, we don’t want to discount the hard work that went into the study and the fact that our diligence has paid off," said Rietz.

Having so far accomplished their exploitation objectives using the simulation, McCollum and his group are considering expanding the use of models. "Because of our modeling success, Nuevo is excited that it may be able to apply future Ryder Scott simulations to other larger, undeveloped fields," he said.

[About] [Calgary Office] [Client List] [Contact] [Downloads] [Financial Institutions]
[International Experience] [Newsletter] [Services] [What's New] [Home]