Abstract

The water softening process is integral to oil production operations, particularly in fields employing cyclic steam injection for enhanced oil recovery (EOR). This analysis examines a dataset spanning from 2022 to 2024, encompassing variables such as date, sample type, oil and water content (O.W), total suspended solids (TSS), pH, chloride (Cl-), conductance (μΩ/cm), and calcium carbonate (CaCO3). Despite significant missing values, the subset focusing on “OUT SUAVIZADA” samples reveals fewer gaps for pH and none for Cl and conductance. However, CaCO3 presents notable missing data. A time series plot reveals variability and potential seasonal patterns in key variables, with CaCO3 and conductance displaying positive correlation. Further analysis via scatterplot matrix and boxplots highlights weak correlations and overlapping distributions among variables. Recommendations include enhancing process control, investigating seasonal trends, and improving data quality management. Leveraging advanced analytical techniques and collaborating with domain experts can optimize the water softening process and support EOR initiatives.

Keywords: Water softening process, Enhanced oil recovery (EOR), Oil production operations, Data analysis, Time series analysis, Process variability, Missing data, Correlation analysis, Process optimization, Data quality management.

Introduction

The water softening process is a critical component in oil production operations, particularly in fields utilizing cyclic steam injection for enhanced oil recovery (EOR). The process involves treating raw water to remove hardness-causing minerals, such as calcium and magnesium ions, to produce softened water suitable for steam generation. Ensuring consistent water quality and operational efficiency is crucial for protecting downstream equipment and maximizing oil production.

Objetive

To conduct a comprehensive performance baseline analysis of water softening processes in the context of enhanced oil recovery (EOR), utilizing data-driven approaches to identify opportunities for optimization and improve operational efficiency in oil production operations.

Specifics

  1. Baseline Assessment: Conduct a comprehensive baseline assessment of existing water softening processes in oil production operations to establish a foundation for optimization efforts.

  2. Performance Evaluation: Evaluate the performance of current water softening methods in terms of their effectiveness in reducing water hardness and minimizing scaling potential in oil extraction systems.

  3. Data Compilation and Analysis: Compile and analyze historical data on water quality parameters, operational variables, and oil recovery rates to identify trends, correlations, and potential areas for improvement.

  4. Identification of Bottlenecks: Identify operational bottlenecks and challenges associated with the water softening process, such as equipment limitations, chemical dosing inefficiencies, and maintenance issues.

  5. Recommendation Development: Develop targeted recommendations for optimizing the water softening process, including adjustments to treatment protocols, equipment upgrades, and procedural enhancements.

Scope

  1. Oil Recovery Context: The analysis is focused specifically on the water softening process within the context of enhanced oil recovery (EOR) operations, considering its critical role in maintaining operational efficiency and maximizing oil extraction rates.

  2. Data-Driven Insights: Emphasis is placed on leveraging historical data and empirical evidence to inform decision-making and guide optimization strategies, ensuring that recommendations are grounded in real-world observations and outcomes.

  3. Practical Implementation: Recommendations are tailored to facilitate practical implementation within existing operational frameworks, with a focus on achievable short-term gains and long-term sustainability.

  4. Continuous Improvement Perspective: The analysis acknowledges the iterative nature of process optimization and encourages ongoing monitoring, feedback, and refinement to sustain improvements over time.

Process overview

Components of the process

  1. Raw Water Source: Raw water is drawn from production wells within the oil field, characterized by parameters such as Total Dissolved Solids (TDS) concentration, Total Hardness, pH, Conductance, and Turbidity.

  2. Filtration Stages: The raw water undergoes several filtration stages aimed at removing suspended solids, organic matter, and certain dissolved solids. Filtration media, including sand, anthracite, and activated carbon.

  3. Ion Exchange Softening: Cation exchange resins play a pivotal role in the process by selectively removing calcium and magnesium ions from the water, substituting them with sodium or potassium ions. Upon depletion, the resin undergoes a regeneration sequence using sodium chloride solution.

  4. Storage and Distribution: Softened water is stored in a designated tank equipped with level measurement and control systems to thwart oxidation. Transfer pumps then distribute softened water to steam generators situated across various zones within the field.

Importance of the water softening process

  1. Enhanced Oil Recovery (EOR): Softened water is crucial for cyclic steam injection, an EOR method that improves crude oil production by reducing viscosity and increasing well productivity.

  2. Equipment Protection: The water softening process protects downstream equipment, such as steam generators, from scale formation and corrosion.

  3. Operational Efficiency: By ensuring consistent water quality, the process contributes to the reliable and efficient operation of steam injection systems.

The water softening process plays a vital role in oil production operations, particularly in fields utilizing cyclic steam injection for enhanced oil recovery. Understanding its components and importance is essential for optimizing production processes and ensuring the longevity of equipment.

Moriche´s process description

  1. Utilization of Softened Water: The softened water is used for cyclic steam injection, an enhanced oil recovery method that injects wet steam directly into the producing formation to reduce crude oil viscosity and increase well production.

  2. System Capacity: The water softening system at Campo Moriche generates and stores industrial and softened water with optimal parameters for steam generation. It consists of several softening plants with a total design capacity of 1,300 gpm (1,040 gpm at 80% efficiency).

  3. Feed Water Characteristics: The feed water is raw water from several production wells with a total dissolved solids (TDS) concentration of 184 ppm, total hardness of 20 ppm CaCO3, pH of 8.28, conductance of 382 (μΩ/cm) and turbidity of <1 NTU.

  4. Treatment Process: The water treatment process involves filtration stages using sand, anthracite, and activated carbon to remove suspended solids, organic matter, and some TDS and dissolved solids. This protects the ion exchange resin and increases its lifespan.

  5. Softening Process: Cation exchange resins are used to remove calcium and magnesium ions, replacing them with sodium or potassium ions. After the resin is exhausted, a regeneration sequence is performed using sodium chloride solution.

  6. Storage and Distribution: The softened water is stored in a 10,000 bbl storage tank (TK-ETR-003) with level measurement, high/low level switches, and a gas blanketing system to prevent oxidation.

  7. Control System: The control system monitors and controls the operation of the softening system using alarm values based on instrumentation readings.

  8. Water Transfer: The softened water is transferred to the steam generators using a set of six pumps, including two booster pumps (P-455-01A/B) and four distribution pumps (P-ETR-002A/B/C/D). The flow is distributed to different zones of the field and controlled by a flow control loop (FIC-4501) and a control valve (FV-4501).

Data Exploration

Categorical variables

The column “muestra” is a categorical variable from the dataset containing names/labels for different sampling sites or sources related to treatment in the Moriche treatment plants. The analysis consists in identifying the unique values and possible duplicates or similar entries. There are 71 unique values/levels in this variable:

##                         muestra
## 1                              
## 2       ECOTECNIA (Out TK-W-01)
## 3       ECOTECNIA (Out TK-W-02)
## 4       ECOTECNIA (Out TK-W-03)
## 5      ECOTECNIA (Out TK-WT-01)
## 6      ECOTECNIA (Out TK-WT-02)
## 7                101-A (ABARCO)
## 8                102-A (ABARCO)
## 9            AGUA CRUDA DE POZO
## 10  AGUA DOMINION (Out TK-WT03)
## 11 AGUA ECOTECNIA (Out TK-WT01)
## 12        AGUA FILTRADA (GRIFO)
## 13        AGUA FILTRADA DEPURAR
## 14       AGUA FILTRADA EDOSPINA
## 15                       BOMBEO
## 16                   FWKO 250 B
## 17                  GENERADORES
## 18        IN CAPTACION DEPURAR.
## 19       IN CAPTACION EDOSPINA.
## 20                       IN FIL
## 21                    IN FILTRO
## 22                   IN FILTRO 
## 23                  IN FILTRO A
## 24                  IN FILTRO C
## 25                        IN SK
## 26           OUT 101-A (ABARCO)
## 27   OUT 101-A / 102-A (ABARCO)
## 28           OUT 102-A (ABARCO)
## 29                      OUT FIL
## 30                   OUT FILTRO
## 31                  OUT FILTRO 
## 32                 OUT FILTRO A
## 33                      OUT MTB
## 34                OUT SUAVIZADA
## 35                      OUT TTO
## 36                  OUT-TKWT-03
## 37                P DEPURAR # 1
## 38                P DEPURAR # 2
## 39                P DEPURAR # 3
## 40           PLANTA 101 EIS ABA
## 41           PLANTA 102 EIS ABA
## 42             PLANTA DEP 3 B-1
## 43             PLANTA DEP 3 B-2
## 44             PLANTA DEPURAR 1
## 45             PLANTA DEPURAR 2
## 46             PLANTA DEPURAR 3
## 47            PLANTA DEPURAR 3 
## 48   PLANTA DEPURAR 3 - BANCO-1
## 49   PLANTA DEPURAR 3 - BANCO-2
## 50         PLANTA DEPURAR 3 B-1
## 51          PLANTA DEPURAR 3 B2
## 52     PLANTA DEPURAR 3 BANCO 1
## 53     PLANTA DEPURAR 3 BANCO 2
## 54           PLANTA EDOSPINA  2
## 55            PLANTA EDOSPINA 1
## 56            PLANTA EDOSPINA 2
## 57          POCETA (ABARCO) PH2
## 58          TK WT 03 (DOMINION)
## 59                     TK WT-03
## 60            TK-103 (DOMINION)
## 61           TK-3-WT (DOMINION)
## 62                      TK-5-02
## 63         TK-503-WT (DOMINION)
## 64                      TK-7-03
## 65                       TK-W03
## 66         TK-WT-01 (ECOTECNIA)
## 67                     TK-WT-02
## 68         TK-WT-02 (ECOTECNIA)
## 69                    TK-WT-503
## 70           TK-WT03 (DOMINION)
## 71                     TK-WT503

Some values appear to be direct duplicates:

  • “PLANTA DEPURAR 3” and “PLANTA DEPURAR 3” (differs by a trailing space)
  • “PLANTA DEPURAR 3 - BANCO-1” and “PLANTA DEPURAR 3 BANCO 1”
  • “PLANTA DEPURAR 3 - BANCO-2” and “PLANTA DEPURAR 3 BANCO 2”
  • “PLANTA DEPURAR 3 B-1” and “PLANTA DEP 3 B-1”
  • “PLANTA DEPURAR 3 B2” and “PLANTA DEP 3 B-2”

Some values are very similar but have slight variations:

  • “TK-WT-01 (ECOTECNIA)” and “AGUA ECOTECNIA (Out TK-WT01)”
  • “TK-WT03 (DOMINION)” and “TK WT 03 (DOMINION)” and “AGUA DOMINION (Out TK-WT03)”
  • Multiple entries for “ABARCO” with different codes like 101-A, 102-A

There seem to be groups of values related to:

  • ECOTECNIA

  • DEPURAR plants 1/2/3

  • EDOSPINA plants 1/2

  • ABARCO 101/102 plants

  • DOMINION

  • Specific tank codes like TK-WT-03

Raw data overview

##       id                fecha               datetime                     
##  Length:28032       Min.   :2022-02-10   Min.   :2022-02-10 03:00:00.00  
##  Class :character   1st Qu.:2022-10-07   1st Qu.:2022-09-24 08:40:00.00  
##  Mode  :character   Median :2023-04-15   Median :2023-04-07 21:30:00.00  
##                     Mean   :2023-04-10   Mean   :2023-04-03 16:50:05.32  
##                     3rd Qu.:2023-10-15   3rd Qu.:2023-10-10 15:00:00.00  
##                     Max.   :2024-04-21   Max.   :2024-04-21 17:00:00.00  
##                     NA's   :2417         NA's   :5050                    
##           muestra           o.w               tss                ph         
##  BOMBEO       : 2346   Min.   :    0.0   Min.   :    0.0   Min.   :   4.00  
##  IN SK        : 2346   1st Qu.:    8.0   1st Qu.:   69.0   1st Qu.:   8.00  
##  OUT FILTRO   : 2341   Median :   22.0   Median :  196.5   Median :   8.00  
##  OUT MTB      : 2327   Mean   :  948.7   Mean   :  411.7   Mean   :  14.51  
##  OUT SUAVIZADA: 2316   3rd Qu.:  610.0   3rd Qu.:  674.8   3rd Qu.:   9.00  
##  IN FILTRO    : 2220   Max.   :71400.0   Max.   :18860.0   Max.   :1708.00  
##  (Other)      :14136   NA's   :16195     NA's   :16194     NA's   :24811    
##        Cl           conductance         CaCO3                Fe        
##  Min.   :   1.00   Min.   :   0.0   Min.   :   0.000   Min.   : 0.000  
##  1st Qu.:  15.00   1st Qu.: 152.0   1st Qu.:   0.000   1st Qu.: 1.000  
##  Median :  39.00   Median : 190.0   Median :   0.000   Median : 2.000  
##  Mean   :  69.03   Mean   : 495.3   Mean   :   1.076   Mean   : 2.598  
##  3rd Qu.:  75.00   3rd Qu.: 258.0   3rd Qu.:   0.000   3rd Qu.: 3.000  
##  Max.   :1873.00   Max.   :2068.0   Max.   :1730.000   Max.   :92.000  
##  NA's   :24843     NA's   :25172    NA's   :16288      NA's   :15015   
##        Ba             SO4           analista         observaciones     
##  Min.   :0.000   Min.   : 0.000   Length:28032       Length:28032      
##  1st Qu.:2.000   1st Qu.: 0.000   Class :character   Class :character  
##  Median :4.000   Median : 0.000   Mode  :character   Mode  :character  
##  Mean   :4.161   Mean   : 2.534                                        
##  3rd Qu.:6.000   3rd Qu.: 1.000                                        
##  Max.   :9.000   Max.   :93.000                                        
##  NA's   :26062   NA's   :27901

The dataset contains 28,032 observations spanning from February 2022 to April 2024, with variables related to the water treatment process, such as date, sample type, oil and water content, total suspended solids, pH, chloride, conductance, calcium carbonate, iron, barium, and sulfate. The dataset also includes analyst and observation information. The sample type variable has various levels, indicating different sampling points or stages in the treatment process. Most of the numeric variables have missing values, with some variables having a significant number of missing entries.

Sampling Points: The presence of various levels in the “muestra” (sample type) variable suggests that the dataset includes observations from different stages or sampling points in the water treatment process. This could allow for the analysis of process performance at different stages and identify potential bottlenecks or areas for improvement.

Process of interest - ‘muestra’ = “OUT SUAVIZADA”

NA’s based-in criterial

Se identifican las variables de interés para el proceso de suavizado:

##   Cl pH   ow  tss CaCO3 Fe  Ba  SO4 Condunctance
## 1 49 70 2314 2311  2080 74 369 2195          218

Con base en los NA’s se infiere que CaCO3, SO4, OW, TSS no son variables de interés para el proceso de suavizado, por lo tanto se omitirán del dataset a analizar.

Dataset description

##   fecha Cl pH Fe  Ba Condunctance
## 1    35 49 70 74 369          218
## [1] 815
  The working dataset is a subset focused on the "OUT SUAVIZADA" (softened water) sample type.
  It contains 2,316 observations with variables like fecha (date), pH, chloride, conductance, Fe and Ba.

Para el data set se determinan 815 NA’s los cuales seran imputados mediante Multivariate Imputation by Chained Equations y los outliers seran identificados con un algoritmo de parendizaje automático Isolation Forest. Los outliers serán convertidos en NA’s para que no se contraiga el dataset y permita que el análisis de series de tiempo fluya. Una vez se tengan todos los NA’s tanto por datos faltantes como por outliers, seran imputados con MICE.

## NULL

Analysis

Relationships description

  1. Neither: No significant correlation expected between the variables based on their nature or process knowledge.

  2. Weak Positive: A slightly positive correlation is expected between the variables, indicating that as one increases, the other tends to increase slightly.

  3. Weak Negative: A slightly negative correlation is expected between the variables, indicating that as one increases, the other tends to decrease slightly.

  4. Moderate Positive: A moderate positive correlation is expected between the variables, suggesting that as one variable increases, the other tends to increase moderately.

  5. Moderate Negative: A moderate negative correlation is expected between the variables, suggesting that as one variable increases, the other tends to decrease moderately.

  6. Strong Positive/Negative: A strong positive or negative correlation is expected between the variables, indicating a strong linear relationship where one variable increases or decreases significantly with the other.

Scatterplot matrix and boxplot analysis

The scatterplot matrixes (Images 1, 2) and boxplot (Image 3) collectively offer valuable insights into the distributions and interrelationships among the key water quality variables.

One notable observation from the scatterplot matrix is the presence of weak linear correlations among the variables. Specifically, the highest correlation coefficient observed is 0.244, which pertains to the relationship between pH and CaCO3. This finding indicates that while some degree of association exists between certain variables, it is not particularly strong.

Additionally, the boxplots depict relatively narrow value ranges for each variable, accompanied by minimal outliers. This suggests that the water softening process effectively maintains these parameters within specific limits, contributing to the overall consistency and stability of water quality.

Furthermore, the density plots within the scatterplot matrix highlight overlapping distributions of the variables. This overlap reinforces the earlier observation of weak correlations, indicating that the variables tend to vary independently of each other rather than exhibiting strong pairwise relationships.

Finding relationships

Variable pH Cl Conductance Fe Ba
pH - Weak Negative Neither Neither Weak Negative
Cl Weak Negative - Weak Positive Neither Neither
Conductance Neither Weak Positive - Neither Moderate Positive
Fe Neither Neither Neither - Neither
Ba Weak Negative Neither Moderate Positive Neither -

Expected relationships

Variable pH Cl Conductance Fe Ba
pH - A weak negative correlation is expected because lower pH values can increase the solubility of chloride salts. Neither Neither Neither
Cl = - A weak positive correlation is expected because chloride ions contribute to the overall conductivity of the water. Neither Neither
Conductance Neither = - Neither A moderate positive correlation is expected because barium ions contribute to the overall conductivity of the water.
Fe Neither Neither Neither - Neither
Ba Neither Neither = Neither - 
## [1] "/Users/john/projects_improve_process/quaility_treatment"
## Loading required package: isotree
## Loading required package: mice
## 
## Attaching package: 'mice'
## The following object is masked from 'package:stats':
## 
##     filter
## The following objects are masked from 'package:base':
## 
##     cbind, rbind
## 
##  iter imp variable
##   1   1  ph  Cl  conductance  Fe  Ba
##   1   2  ph  Cl  conductance  Fe  Ba
##   1   3  ph  Cl  conductance  Fe  Ba
##   1   4  ph  Cl  conductance  Fe  Ba
##   1   5  ph  Cl  conductance  Fe  Ba
##   2   1  ph  Cl  conductance  Fe  Ba
##   2   2  ph  Cl  conductance  Fe  Ba
##   2   3  ph  Cl  conductance  Fe  Ba
##   2   4  ph  Cl  conductance  Fe  Ba
##   2   5  ph  Cl  conductance  Fe  Ba
##   3   1  ph  Cl  conductance  Fe  Ba
##   3   2  ph  Cl  conductance  Fe  Ba
##   3   3  ph  Cl  conductance  Fe  Ba
##   3   4  ph  Cl  conductance  Fe  Ba
##   3   5  ph  Cl  conductance  Fe  Ba
##   4   1  ph  Cl  conductance  Fe  Ba
##   4   2  ph  Cl  conductance  Fe  Ba
##   4   3  ph  Cl  conductance  Fe  Ba
##   4   4  ph  Cl  conductance  Fe  Ba
##   4   5  ph  Cl  conductance  Fe  Ba
##   5   1  ph  Cl  conductance  Fe  Ba
##   5   2  ph  Cl  conductance  Fe  Ba
##   5   3  ph  Cl  conductance  Fe  Ba
##   5   4  ph  Cl  conductance  Fe  Ba
##   5   5  ph  Cl  conductance  Fe  Ba
## Loading required package: zoo
## 
## Attaching package: 'zoo'
## The following objects are masked from 'package:base':
## 
##     as.Date, as.Date.numeric
## # A tibble: 6 × 6
## # Rowwise: 
##   fecha         ph    Cl conductance    Fe    Ba
##   <date>     <dbl> <dbl>       <dbl> <dbl> <dbl>
## 1 2022-02-10     8    71         109 0.584  4.19
## 2 2022-02-10     8    35         106 0.584  4.19
## 3 2022-02-10     8    62         130 0.584  4.19
## 4 2022-02-11     8    58         121 0.584  4.19
## 5 2022-02-11     8    14         118 0.584  4.19
## 6 2022-02-11     8    17         120 0.584  4.19

Conclusions and Recommendations

The preliminary analysis of the water softening process data at Campo Moriche indicates the absence of significant correlations or patterns due to the limited number of observations and the timeframe of the data. However, it’s evident that robust data management practices will be crucial for future improvements in the softening process.

Despite the limitations, this initial assessment provides valuable insights into the importance of gathering, registering, and processing data related to the softening process. It serves as a baseline for future analyses and optimizations, highlighting the need for comprehensive and detailed data collection to drive improvements in the water softening process.

Key recommendations

  1. Process Control and Monitoring: Implement improved process control and monitoring strategies to maintain consistent water quality within desired ranges, addressing the observed variability and potential process excursions.

  2. Seasonal Patterns and Cyclic Trends: Investigate seasonal patterns or periodic trends to identify their underlying causes and implement appropriate mitigation strategies or process adjustments.

  3. Relationship between Variables and Operational Parameters: Analyze the relationships between process variables and other operational parameters (e.g., flow rates, resin regeneration cycles) to reveal opportunities for process optimization and variable control.

  4. Data Quality and Management: Address the issue of missing data, either through imputation or removal of observations, for robust analysis. Additionally, Implement a comprehensive data management system to ensure consistent and accurate data collection, storage, and retrieval.

  5. Advanced Analytical Techniques: Explore advanced analytic techniques, such as machine learning algorithms or multivariate statistical methods, to uncover complex patterns and relationships in the data, once a more complete and high-quality dataset is available.

  6. Collaboration and Domain Expertise: Involve subject matter experts and consider practical considerations and operational constraints to align the analysis with the objectives of the water softening process and provide actionable insights.

Immediate recommendations

  1. Establish clear protocols for data collection, ensuring consistency and accuracy across all relevant variables.

  2. Implement standardized procedures for data entry, validation, and quality assurance to maintain data integrity.

Short-term recommendations

  1. Utilize rigorous statistical methods and visualization tools to extract meaningful insights from the data.

  2. Conduct regular reviews of data collection and analysis processes to identify areas for optimization and incorporate stakeholder feedback.

Medium-term recommendations

  1. Investigate potential process excursions and outliers to enhance process control and monitoring strategies.

  2. Analyze seasonal patterns and cyclic trends to identify underlying causes and implement appropriate mitigation strategies.

  3. Involve subject matter experts and stakeholders to ensure alignment with process objectives and operational constraints.

References

Ecopetrol. (2024). Water Softening Processes Raw DataFrame [GitHub repository]. Retrieved from https://github.com/JOHNRAMIREZ74/Water-Softening-Processes/tree/cd8811ddead1d5b6251105e12a98ddafb6b2551e/data

Ecopetrol. (2023, November 9). Instructivo Operacional Estación de Trasferencia y Recibo ETR Moriche [GitHub repository]. Retrieved from https://github.com/JOHNRAMIREZ74/Water-Softening-Processes/blob/cd8811ddead1d5b6251105e12a98ddafb6b2551e/Instructivo_Operacional_Estaci%C3%B3n_de_Trasferencia_y_Recibo_ETR_Moriche.pdf