The case study used here is the XYZ field development with water injection, where the EPSA IV agreement is used. The water injection operation begins at the start of field production as part of the operator company’s policy to improve reservoir performance. In this scenario, the XYZ field will be supplied by 50 producing wells, with an average daily production flow rate of around 60 thousand barrels of oil per day. Other details are shown in
Table 1. At the 60,000 STB/day rate, the plateau rate will last for 6 years. In addition, four peripheral water injections will be drilled to maintain pressure. Cumulative production is expected to reach approximately 219 million barrels by the end of 2037, with an anticipated recovery factor (RF) of 22% of the original oil in place (OOIP). Because of the field’s relatively high gas–oil ratio (GOR) of 800 SCF/STB and water production, a three-stage train of a three-phase separator is chosen for installation. The field’s condensate production is estimated to be 30 STB/MMSCF. As a result, a gas plant will be required to separate the hydrocarbon byproducts from the produced gas.
Table 1 shows the development expenses of the XYZ field as well as the basic assumptions used to evaluate the project development plans with water injection.
The field was developed using a primary and secondary recovery method with water injection. The noise factors in this paper are oil price, gas price, and liquefied hydrocarbon by product (LHP) price.
2.2. Taguchi Method
The Taguchi method has been used successfully to improve the quality of Japanese products since 1960. However, in this paper, it will be used for the first time in developing the agreements of oil between the state and international companies. It is necessary to obtain the desired performance measures while minimizing the consequences of variation, but without eliminating the causes of this variation (since they are difficult to control). It is well-known that “A” and “B” factors affect the net present values of both parties [
6]. The contribution of this study is to examine the effect of different levels of “A” and “B” factors on the variability of the profits of both parties. The main hypothesis is that there are some levels of control factors (“A” and “B” factors) that reduce the fluctuations in some targeted performance measures. In this study, the causes of variability are the oil price, the liquid hydrocarbon price (LHP), and gas price, which are very dynamic. The end result is a robust method that has the minimum sensitivity to variations in the uncontrollable noise factors (prices). In Taguchi’s parameter design approach, the outcome is influenced by two sorts of factors: control factors and noise factors. Control factors are those that can be easily controlled. Noise factors are those that are difficult, impossible, or expensive to control [
26,
27].
The objective of this research is to find the “A” and “B” values that make the agreement more robust against the fluctuations in the prices of oil, LHP, and gas. Robust agreements reduce the negotiation effort. The “A” and “B” levels are used to determine the net present value of the two parties in the agreement. The eight “A” and “B” factors (A1, A1, A3, A4, B1, B2, B3, and B4) are negotiable. This study can help decision makers while negotiating these values to look for better levels that are more robust against the changes in the prices and therefore to be fair for both sides. The ranges of the values of the control factors are relatively long, and therefore trying all the possible values is very time consuming. Therefore, it is better to choose some levels.
In Taguchi’s approach, the design of experimental techniques, particularly Orthogonal Arrays (OAs), are used to systematically modify and evaluate the varied values of each of the control factors. The columns in the OA represent the factor and its related levels, and each row in the OA represents an experimental run at the provided factor settings [
26,
27]. The eight “A” and “B” factors are the control factors. The experimental designer is responsible for determining the proper factor levels for each control factor. In this study, four levels were chosen for each one of the “A” and “B” factors.
Trying out all the combinations will give 65,536 (=4
8) different possible runs. This is a time-consuming calculation. The Taguchi method depends on just trying some of these combinations using the proper OAs. Then, for each one of the runs, the noise combinations are tried. In this study, the noise factors are the changes in the oil price, LHP price, and gas price. Two levels were chosen for each of these noise factors: one for higher price and another one for lower prices. Some ranges of prices were chosen as shown in the lower part of
Table 4. L8 represents the full OA for the noise factors in this study.
Table 4 shows the different levels for both the control and noise factors.
L256 orthogonal design was chosen for control factors. This means that there were 256 different combinations of the levels of the control factor. Some of these runs were deleted because B3 and B4 have a similar level (0.55) in these runs. The number of remaining experimental runs was 224. As mentioned before, L8 was chosen for noise factors. R software was used to make the needed calculations.
Taguchi recommends using an “inner array” and “outer array” approach to build robust design. The “inner array” is the OA containing the control factor settings; the “outer array” is the OA containing the noise factors and their settings that are being investigated. The inner array and outer array are combined to form the “product array” or full parameter design layout. The product array is used to test various combinations of control factor settings across all noise factors in a systematic manner [
26,
27]. The total set of experiments is attained by combining the L224 array of control factors with the L8 array of noise factors. The total number of experiments is the product of the number of runs of outer and inner arrays (224 × 8 = 1792 experiments).
Table 5 shows the first 10 runs (control factors together with the noise factors). Results should be in the lower right part of the figure. Each run has eight experiments. For each run, the average, standard deviation, and signal-to-noise ratio were found. Factor levels that increase the SN ratio are the best. Larger SN ratios mean lower variability.
The SN ratio can be found as follows in Equation (12), where the aim is to keep the SN values around a desired target [
26,
27].
The higher the ratio, the lower the variability due to the noise factors (price fluctuations). The decision makers in the two parties will focus not only on reducing the fluctuations but also on the best profit for each party. The SN ratios must be computed for each run. The SN ratios and average responses can be plotted for each factor against each of its levels. The graphs are tested to choose the factor level which maximizes SN ratio and brings the mean on target. Generally, factors can affect both the variation and the average performance (just like the factors in this study) or one of them (average or variation), or do not affect either the variance or the average. Generally, reducing the variations in the system will make it more robust.
There are two response variables chosen in this study which are the average value of the second-party percent share of production (ASPS) and the percent of profit of the second party (SPP), which can be called “company take”. The percent of the profit of the first party (government take) is simply the same as the SPP. The second party percent share of production usually fluctuates between zero and the maximum second party share. Any small difference in this share makes a difference of millions of USD for both parties. Its value in the first few years is zero. The average value of the nonzero values is chosen to be the response variable, which is the ASPS. It is important to notice that two possible runs with the same ASPS does not guarantee that both of them have the same NPV. Therefore, SPP might be more interesting for the international company.