- First article on the probability of spills associated with a UK shale gas industry
- Spills occur on the well pads and during fluid transportation to and from the site.
- A spill during transportation is predicted for every 19 well pads developed.
- A spill onsite is predicted for every 16 well pads developed.
S.A. Clancya, F. Worralla, R.J.Daviesb, J.G. Gluyasa
aDepartment of Earth Sciences, Durham University, Durham DH1 3LE, UK. bSchool of Natural and Environmental Sciences, Newcastle University, Newcastle NE1 7RU, UK
Received 13 December 2017 Accepted 18 January 2018
© 2018 The Authors
Rapid growth of hydraulic fracturing for shale gas within the USA and the possibility of shale developments within Europe has created public concern about the risks of spills and leaks associated with the industry. Reports from the Texas Railroad Commission (1999 to 2015) and the Colorado Oil and Gas Commission (2009 to 2015) were used to examine spill rates from oil and gas well pads. Pollution incident records for England and road transport incident data for the UK were examined as an analogue for potential offsite spills associated with transport for a developing shale industry.
The Texas and Colorado spill data shows that the spill rate on the well pads has increased over the recorded time period. The most common spill cause was equipment failure. Within Colorado 33% of the spills recorded were found during well pad remediation and random site inspections. Based on data from the Texas Railroad Commission, a UK shale industry developing well pads with 10 lateral wells would likely experience a spill for every 16 well pads developed.
The same well pad development scenario is estimated to require at least 2856 tanker movements over two years per well pad. Considering this tanker movement estimate with incident and spill frequency data from UK milk tankers, a UK shale industry would likely experience an incident on the road for every 12 well pads developed and a road spill for every 19 well pads developed. Consequently, should a UK shale industry be developed it is important that appropriate mitigation strategies are in place to minimise the risk of spills associated with well pad activities and fluid transportation movements.
The predicted number of incidents and spills that are likely to occur during the transportation of fracking related fluids from the development of a UK shale gas industry.
Increased global demand for energy is driving a rapid increase in the use of hydraulic fracturing and horizontal drilling (Gross et al., 2013; Patterson et al., 2017). Hydraulic fracturing allows for enhanced oil and gas extraction from unconventional formations such as low-permeability shale and source rock (McLaughlin et al., 2016).
The process involves high-volume fluid injection of fracturing fluid into a shale reservoir at a sufficient rate to raise downhole pressure above the fracture pressure of the formation rock, when the shale is pressurised fissures and interconnected fractures are formed enabling greater flow rates of gas into the well (Gregory et al., 2011; Wilson et al., 2017). Once the hydraulic fracturing processes are performed the pressure is relieved and the fracturing fluid returns to the surface through the borehole, the returning fluid is termed flowback fluid (Gregory et al., 2011).
Within the USA, fracturing fluids are typically composed of about 90% water, 9% proppant (e.g. sand), and 0.5–1% chemical additives (McLaughlin et al., 2016; Vidic et al., 2013). Additives are generally delivered to the well site in a concentrated form and stored until they are mixed with the base fluid and proppant and pumped down the production well (USA EPA, 2016). Within the USA additives are often stored in multiple, closed containers and moved around the site in specially designed hoses and tubing (USA EPA, 2016).
Flowback fluid is typically highly saline, reaching five times the salinity of sea water (Gregory et al., 2011). It can also contain high levels of dissolved and suspended solids, heavy metals, fracking chemicals, naturally occurring radioactive materials of varying concentrations and hydrocarbons extracted from the formation (Edminston et al., 2011). The volume of flowback that returns to the surface is variable, with between 10 and 50% of the fracturing fluid returning to the surface (Akob et al., 2015) during the ‘flowback period’ (the first two weeks after hydraulically fracturing the rock) (Howarth et al., 2011).
During the active gas production stage, aqueous and non-aqueous liquid continue to be produced in considerably lower volumes than the fracking and flowback fluids over the lifetime of the well (known as produced water – Gregory et al., 2011). Typically within the USA, flowback water and produced water flow from the well to onsite tanks or pits through a series of pipes or flowlines before being transported offsite via trucks or pipelines for disposal or reuse (USA EPA, 2016). Therefore, for the development and exploitation of shale gas resources there would be three types of potentially polluting liquids to consider: the fracking fluid; the flowback water; and the produced water. In addition undiluted chemical additives also need to be considered.
In addition to the risks associated with transport, as with other outdoor practises, well pad sites (the area required for the borehole, drilling equipment, piping and storage) are exposed to extreme weather and environmental conditions (e.g. heavy rainstorms, severe windstorms, floods and freezing conditions) which can also lead to spills and leaks of potentially hazardous fluids on the well site (Eshleman and Elmore, 2013). Even with appropriately designed storage equipment for additives, blended hydraulic fracturing fluids, flowback fluids and produced water, spills could occur.
Currently there is no shale gas production within Europe, however exploration wells are underway and the public have expressed many concerns regarding the potential for water contamination. Included in the perceived risk to water is the potential for polluting spills and leaks to contaminate land, surface water and groundwater, which if severe may lead to polluted fluid being exposed to humans and natural ecosystems (Eshleman and Elmore, 2013; Vengosh et al., 2014). Based on our review there have been no studies published in the peer-reviewed scientific literature addressing the potential for spills and leaks, either onsite or offsite, from possible hydraulic fracturing sites within Europe.
Within the USA a number of studies have considered the risk to the surface and subsurface environment from spills and leaks. Gross et al. (2013) examined the Colorado Oil and Gas Commission’s database of incidents and found surface spills were associated with <0.5% of the active wells. Drollette et al. (2015) found that groundwater near the Marcellus shale gas operations in north eastern Pennsylvania had been contaminated by diesel-range organic compounds via accidental release of fracturing fluid chemicals, derived from the hydraulic fracturing activities at the surface.
DiGiulio et al. (2011) found leakages from storage and disposal pits were responsible for the high concentrations of benzene, xylenes, gasoline range organics, diesel range organics and total purgeable hydrocarbons found in shallow ground water around the Pavillion field in Wyoming. The USA Environmental Protection Agency (EPA) assessed data from nine state agencies, nine oil and gas production well operators, nine hydraulic fracturing service companies and determined 457 hydraulic fracturing-related spills occurred between January 2006 and April 2012 (USA EPA, 2015). More recently Patterson et al. (2017) considered spills from unconventional oil and gas wells, in Colorado, New Mexico, North Dakota and Pennsylvania from 2005 to 2014, recording that between 2 and 16% of wells reported a spill each year.
These reviews of spills and leaks have only considered onsite incidents, not those occurring offsite. The average multi-stage well in the USA requires hundreds to more than a thousand round trips to transport equipment, chemicals, sand and water required for well development and hydraulic fracturing (Adgate et al., 2014; Muehlenbachs and Krupnick, 2013).
Muehlenbachs and Krupnick (2013) found a significant increase in the total number of accidents and accidents involving heavy trucks in counties with a relatively large degree of shale gas development, compared to those counties with less (or no) development: they found one additional well drilled per month raised the frequency of an accident by approximately 2% and increased the risk of a fatality by 0.6%.
The Texas Department of Transportation also noted that the influx of traffic from the development of the Permian Basin had generated an increase in the number of road traffic accidents: a 27% increase in roadway fatalities, trucks were involved in 7% of these reported crashes (Texas Department of Transportation, 2013). These studies did not consider the potential for spills and leaks from these offsite incidents.
With the possibility of a shale gas industry emerging within the UK Goodman et al. (2016) determined the number of truck visits required over the lifetime of a single-well pad with six-wells, and from this, the impact upon local air quality, greenhouse gas emissions and noise emissions. However the number of incidents and spillages were not considered. Lacey and Cole (2003) used information from UK databases on vehicular flow of tankers, accident rate and the probability that an accident would result in a spill; from this they predicted the expected number of spills per year. Their analysis predict the likelihood of a spill which exceeds 150 kg of chemical load spilling on a 2 km section of road is once in 370 years, with a range of 75 to 1800 years.
Thus, the aim of this study was to assess the probability of surface spills and leaks of undiluted additives, fracturing fluid, flowback water and produced water, assessing the probability of spills occurring both onsite (on the well pad) and offsite (during fluid transportation). Secondly we have assessed volumes spilt and the underlying cause of spills in analogue developments to help generate mitigation strategies for potential future sites in the UK.
Approach and methodology
A leak is a way for fluid to escape a container or fluid-containing system. The word leak usually refers to a gradual loss; whilst a sudden loss is usually called a spill. For simplicity this study refers to any accidental and undesired escape of fluid as a spill. Additionally we have not distinguished between the different types of fluids spilt (e.g. flowback water, fracking fluid, produced waters), we are aware that the toxicity of the type of fluid spilt and therefore the impact of the spill can vary considerably, for example spilling a highly saline flowback water is very different to spilling produced waters contaminated with BTEX or crude oil. However, within this study we have focused on the probability of an incident occurring rather than the consequence.
Without a shale gas industry currently operating within Europe information has been drawn from both onsite and offsite experiences in the USA and analogues from within the UK. Due to differences in the source and occurrence of the spills this study has analysed onsite and offsite incidents separately. Two USA state data sources were considered: the Texas Railroad Commission (Texas RRC – RRC, 2017a) and the Colorado Oil and Gas Commission (COGCC – COGCC, 2017a, COGCC, 2017b).
The recorded spills have been evaluated to assess the type, volume and reasons for the currently occurring spills. From this spill analysis the probability of spills onsite for potential shale gas developments within Europe has been assessed. In England, spills from oil and gas sites are reported to the Environment Agency (EA) and recorded in the pollution incident database. This was analysed to access the number of incidents that have occurred on conventional well pads within England.
Without a fully developed shale industry within the UK, fracking fluid, produced water and flowback fluid will be transport to and from the site via tanker trucks. It is currently believed the new development at the Preston New Road site in Lancashire will require produced and flowback fluid to be transported over 80 km by truck to the Davyhulme wastewater treatment works in Manchester. With increased transportation from Preston New Road to Davyhulme there is an increased chance of an incident or spill offsite.
With a lack of information in Europe and the USA for incidents offsite, UK milk and fuel (petrol and diesel) tanker incidents were analysed as an analogue to determine the probability of an incident related to hydraulic fracturing occurring on the road for different shale gas development scenarios. These vehicle types have been identified within the records and were considered a good analogue for the transport required within a UK shale industry, as they often operate on rural roads carrying liquid that is a pollutant with respect to surface waters. Recorded tanker incidents have been cross-correlated with the pollution incident database for England to determine their environmental impact.
Texas Railroad Commission (Texas RRC) database
The Texas RRC enforces the delineation and reporting of any spill of 0.8 m3 or more within the state of Texas (RRC, 2017b). The dataset includes surface spills of crude oil, gas well liquid1, products2 and combined3 (RRC, 2017a). This data is publically available and documents the number of spills, volume spilt, spill type, facility type the loss was from and the cause for all spills from 2009 to 2016. The data indicates the gross loss per spill, the amount of spill recovered and the net loss.
The data were evaluated for each year individually and then compiled to assess trends within the whole dataset. The statistical significance of trends was assessed using a t-test and in all cases significance was judged at a probability of not being zero of 95%. The Texas RRC also records the number of wells active per year and the volumes of crude oil produced; from these the percentage of produced crude oil spilt was calculated. From the average number of spills per year and the average number of active wells per year, the average number of spills per well has also been calculated.
Colorado Oil and Gas Commission (COGCC) database
The COGCC require operators to fully report: (1) spills of any size that impact or threaten to impact waters of the state (streams, lakes, ponds, drainage ditches), structures, livestock, public byways; (2) spills >0.2 m3 that released exploration and production (E&P), or produced water outside of the berm or other secondary containment; (3) spill of 0.8 m3 or more, regardless of whether the spill was contained within the berms or other secondary containments (COGCC, 2015). The COGCC has two spill databases, due to considerable changes in processing and data collection these databases are not comparable and have been analysed separately, these two datasets are henceforward referred to as: “1999–2015 spill data”; and “2014–2015 spill data” (COGCC, 2017a, COGCC, 2017b).
Both datasets included data for 2016; however, data were only available for the first two quarters of 2016, as the dataset for 2016 was incomplete it was not included in this study. The “2014–2015 spill dataset” provides the following information on each spill; timing, location, type and volume, facility type (where breach occurred) and the impact on land and surrounding environment.
Conversely the “1999–2015 spill dataset” is less comprehensive, only consisting of the number of active wells, the annual volume of oil and water spilt and produced and the percentage of the produced oil and water spilt. From the “1999–2015 spill data” we have assessed the changes and patterns in oil and water spill numbers and volumes over the 17 years recorded. Using the number of active wells per year and the “1999–2015 spill data” the average number of spills per well has been calculated.
Pollution incident database
The Environment Agency records the pollution incidents in England and classifies them according to their impact on the population, the environment and level of response required (EA, 2017). Each incident is recorded by date and location and is categorised on pollution type and impact. The pollution impact category system is 1 (major) to 4 (no impact). The pollution incident database for England contains 12,335 incidents recorded between March 2001 and December 2016 (EA, 2017).
To determine the number and cause of incidents related to well integrity failure within England, Davies et al. (2014) analysed the pollution incident database. Davies et al. (2014) only reported incidents which could be confirmed as being due to well integrity failure, whereas as this study considered all incidents reported, from any well pad. Identification and analysis of the cause of releases in currently operating industries allows for lessons to be learnt and mitigation strategies to be put in place avoiding repeating these incidents.
Onsite industrial development scenarios
The UK’s Institute of Directors (IoD) have suggested several shale gas development scenarios for the UK, the first is based on the development of a 10-well pad of 10 laterals (one well pad with 10 wells each with one lateral) (Taylor et al., 2013). The second involved the development of a 10-well pad of 40 laterals (one well pad with 10 well each with four laterals) (Taylor et al., 2013).
Without a shale gas industry currently operating within Europe, this study has used an analogue of UK milk tanker journeys to predict the probability of spills during the transportation of fracking fluid, produced water and flowback water to and from the well site. Within this study milk tankers are defined as vessels used to transport large quantities (approximately 30 m3) of milk. This study does not include references to milk floats, vans or lorries. Assuming an average milk tanker size of 30 m3, some 366,667 milk tanker journeys are required to transport the 11 million m3 of milk produced by British farmers each year (Taylor et al., 2013).
A search for local media reports involving milk tanker incidents in the UK between 1998 and 2016 was carried out. The data collection involved searching for all online media articles that mentioned “milk”, “tanker”, “accident”, “incident”, “road”, “crashes”, “overturned”, “UK” and “spillage”. There was no discrimination on the type of report or article, authorship or publisher used. Incidents due to engine fires were not recorded. The number of milk tanker incidents that were reported was recorded; those that resulted in a spillage of milk or flammable liquid (e.g. diesel) were logged; as were the volumes spilt if documented and cause of incident. If the incident resulted in injuries or fatalities these were noted.
Where possible incidents reported in the media were matched to those recorded in the pollution incident database, and the type and scale of the pollution caused by the spill incidents assessed. As this database only includes incidents from England, only these have been matched.
The UK road fuel (petrol and diesel) tanker fleet is estimated to be around 1000–1500 vehicles, these are estimated to travel some 220,000 km each year (Robinson et al., 2014). The size and volume capacity of fuel tanker trucks varies considerably. Commonly large tanker trucks with capacities of 21–44 m3 are used to transport petrol and diesel to filling stations (Madigan, 2017). Fuel tankers have been studied, as similarly to milk tankers they are a good analogue for the truck movements that would be associated with a UK shale industry.
Unlike milk tankers the average number of journeys required each year to transport the nations fuel is undocumented in the literature. Therefore, the number of fuel tanker journeys has been determined from the known volume of motor fuel consumed by the UK, recorded by the Department for Business, Energy and Industrial Strategy (BEIS, 2017) and the average tanker size used. This estimate was then used to determine the probability of an incident or spill per year.
The Transport Research Laboratory (TRL) compiled data on all tanker accidents by carrying out a search of local BBC news reports involving tanker incidents which occurred in the UK between 2009 and 2014 (Robinson et al., 2014). Their data collection involved searching for all media articles that mentioned “tanker” and “accident” on the BBC news website. These were then assessed on whether a spill occurred; a flammable liquid was spilt; an injury resulted; the incident was caused by a collision or the tanker overturned; and if the tanker overturning led to a spillage (Robinson et al., 2014).
This study continued the search for 2015 and 2016 in the same manner to the method used by TRL. In a similar manner to milk tankers a broader search was then conducted just for fuel tankers between 2009 and 2016, using the same search terms to check for milk tanker accidents, with the addition of “milk” being substituted for “fuel”, “petrol” and “diesel”. As with milk tankers, where possible, spill incidents related to fuel tankers were matched to incidents recorded in the pollution incident database.
Offsite industrial development scenarios
The development scenarios proposed by Taylor et al. (2013), along with the annual number of incidents and spills per milk and fuel tanker on the road for 2016 have been used to calculate the potential number of incidents and spills related to a future UK shale industry. Data for 2016 was used to generate the following scenarios as this was deemed the most accurate, being the most up to date.
Taylor et al. (2013) first scenario, based on the development of a single 10-well pad of 10 laterals, would potentially produce 0.9 km3 of gas, requiring 136,000 m3 of water per well. Initially it is likely that the water will be trucked to the site rather than piped, thus requiring between 2856 and 7890 tankers over a 20 year period (Taylor et al., 2013). If tanker movement was concentrated in the early years of drilling activity, which is most likely, this would average out at 3.9–10.8 tanker movements per day over two years, or if spread over 20 years decreasing to 0.4–1.1 per day (Taylor et al., 2013).
The second scenario, based on a single 10-well pad of 40 laterals, potentially producing 3.6 km3 of gas and using 544,000 m3 of water per pad equates to between 11,155 and 31,288 tanker movements over 20 years, or 1.5–4.2 tanker movements per day (Taylor et al., 2013): when averaged out over five years this equals 6.1–17.1 truck movements per day (Taylor et al., 2013). As it is unlikely that just one well pad would be developed the probability of an incident or spill occurring from a number of well pads was estimated.
The analysis of the results has been split into incidents that occurred onsite and incidents that occur offsite, during transportation.
Texas Railroad Commission
The number of reported spills between 2009 and 2015 has increased each year with 675 reported in 2009 and 1485 in 2015 (Table 1). Over the same period the number of producing wells has also increased from 157,807 in 2009 to 193,807 in 2015. The number of spills per producing well increased at an average rate of 0.0006 spills/yr2, the t-test showed that the increase in spill rate is significant at 95% probability.
Of the 7820 spills recorded during the study period the majority (83%) involved the loss of crude oil (Table 1). The most common cause of leakage was due to equipment failure; the second was due to corrosion (rust) of equipment, followed by ‘Acts of God’ and human error. The most common location for a spill to occur was around the tank battery (70% of the spills), followed by the flow line (10%) and pipe line (8%).
Table 1. The annual number of active wells, and associated gross loss, fluid recovered, net loss and percentage recovered for crude oil, gas well liquids or associated products. Data recorded by the Texas Railroad Commission (RRC, 2017a, RRC, 2017c).
The number of crude oil spills has increased year on year since 2009, with 549 crude oil spills reported in 2009 and 1270 in 2015. The average number of crude oil spills per year was 924. The number of crude oil spills per producing well increased at a rate of 0.0001 spills/yr2 – significant at the 95% probability (Table 1). The total annual volume of crude oil spilt varied from 6713 m3 (2009) to 14,158 m3 (2015) (Table 1).
The average rate of change over this seven year period was 805 m3/yr2 which is statistically significant at the 95% probability. Clean-up operations recover some of the lost fluid, however much is left unrecovered. Annually between 50 and 76% of the crude oil spilt is recovered, with an annual average of 59% (Table 1). The largest spill was recorded in 2010 with 3975 m3 of crude oil escaping in one incident; however, 99.7% of this was recovered (Table 1). The largest reported net loss of crude oil for a single spill was 1069 m3 (Table 1).
Between 2009 and 2015, 715 producing wells reported gas well liquid spills. The number of spills involving gas well liquid decreased over the time period analysed, this trend was not statistically significant (Table 1). The total annual volume of gas well liquid spilt ranged from 489 m3 in 2015 to 2438 m3 in 2013 (Table 1). The annual average percentage of gas well liquid recovered was 30% of the amount spilt.
The number of spills involving product varied year on year, from five spills in 2015 to 95 in 2013 (Table 1). Although there has been an increase in the number of wells and spills per year the trend was not statistically significant. The annual percentage recovery rates show that 65% of the product is recovered after a spill (Table 1). The annual average minimum and maximum recovery ranged from 16% in 2012 to 94% in 2010 (Table 1).
For the loss of combined liquids there has been a statistically significant trend over the period of record: in 2009 three cases were recorded, whilst in 2015 154 cases were recorded (Table 1). The annual average minimum and maximum recovery ranges from 20% in 2011 to 91% in 2010 (Table 1).
Colorado Oil and Gas Commission
The 1999–2015 spill data does not distinguish between whether it was oil or water being spilt. It records a total of 6617 spills, the maximum and minimum numbers of spills per year were 789 in 2014 and 193 in 2002 (Table 2). The average number of spills per year was 389. Between 1999 and 2015 there has been an increase in the number of active producing wells and the number of spills at a rate of 0.00017 spills/yr2, however, this increase is not statistically significant. A total of 0.11 km3 of oil and 0.88 km3 of water were produced between 1999 and 2015. Of this 8670 m3 of oil and 81,200 m3 of water were spilt, equivalent to 0.008% and 0.009% of the oil and water produced (Table 2). For this dataset there is no information on recovery rate or reasons for the spills.
Table 2. The annual number of active wells, number of spills and volumes of oil and water produced and spilt for Colorado. Data recorded by the Colorado Oil and Gas Commission (COGCC, 2017a).
Of the 2009–2015 spill data, only years 2014 and 2015 are complete, therefore only those years have been studied. Of the 2893 spills recorded during this period; 563 were oil, 401 condensate, 50 flowback water, 1399 produced water, 78 E&P waste and 129 drilling fluid (Table 3). The volume spilt varies considerably; 188 spills were recorded between >0 and <0.2 m3, 1201 were between ≥0.2 m3 and <0.8 m3, 1051 were between ≥0.8 m3 and <16 m3 and 180 were ≥16 m3 (Table 3). The average length and width of a spill was 33 m and 10 m, whilst the maximum was 1416 m and 152 m.
The average depth to groundwater in the spill locality was 28 m and the average depth the spill impacted was 2.5 m, with a maximum depth impact of 22 m. Just over 73% (2112) of spills had ≥0.16 m3 of fluid leak outside the berm of the well pad, with three sites requiring an emergency pit to be constructed. The average volume of soil that needed to be excavated due to pollution from a spill was 220 m3, with a maximum of 10,780 m3 being removed from one site. Polluted soil was excavated offsite from 471 sites; 62 sites treated the soil onsite, whilst 74 sites had the soil disposed of by alternative methods.
The average volume of groundwater removed was 42 m3, with 484 m3 being the maximum quantity removed from one site. At two sites 1 m3 and 6 m3 of surface water was removed. Of the spills documented; 1107 impacted soil, 260 groundwater, 16 surface water and 30 dry drainage features.
Table 3. The type of fluid and volume of fluid spilt for 2014 and 2015 in the State of Colorado. Data from Colorado Oil and Gas Commission (COGCC, 2017b).
Of the spills 1946 were termed ‘recent’, thus recent or ongoing at the time of discovery. Whereas 947 were termed ‘historical’, therefore the spill occurred at a time unknown or was discovered during activities such as plugging and abandonment or site reclamation. Of the spills 653 were reportedly due to equipment failure, 254 human errors, 186 were historical and 46 were recorded as “other”. Examples of “other” include weather, vandalism and external sources of interference such as cattle.
In 2014, one instance involved cattle rubbing against the valve handle of the wellhead partially opening the valve allowing produced water to spill out. In 2015, there was a report of wild horses pushing open a 2.5 cm valve, this was determined by tracks and faeces left in the area. The most common location facility type from which spills originated from was the tank battery, with 36% of spills initiating there, whereas 6% of the spills were associated with pipelines.
Pollution incident database
Based on data provided by DECC, Davies et al. (2014) comments that there were 143 onshore oil and gas wells producing at the start of the year 2000. Between 2000 and 2013 the Environment Agency recorded nine pollution incidents involving the release of crude oil within 1 km of an oil and gas well. Two of the spills were recorded at the Singleton Oil Field and were caused by borehole cement failure. The other seven pollution incidents were due to leaks from pipework linked to the well (Davies et al., 2014). Between 2000 and 2013 the pollution incident rate was 0.0045 incidents/well/yr.
Using data from the Texas RRC and values from the first scenario (based upon well pads with 10 laterals) the probability of a spill occurring on a developed site in the UK was calculated at 0.06 spills/well pad; therefore there would likely be a spill onsite for every 16 well pads developed (Fig. 1). When the COGCC 1999–2015 spill data and values from the first scenario are used the likelihood of a spill is 0.11 spills/well pad; therefore a spill would likely occur for every 10 well pads developed (Fig. 1).
Fig. 1. The number of sites that need to be developed before a spill is likely to occur onsite, based on data from the Texas RRC and COGCC. Scenario 1 (dark blue bars): Single-well pad with 10-wells, each well is a lateral; Scenario 2 (light blue bars): Single-well pad with 10-wells, each well has four laterals. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Using data from the Texas RRC and values from the second scenario (based upon well pads with 10-wells, each with four laterals) the likelihood of a spill onsite was 0.26 spills/well pad; therefore it is likely that a spill would occur for every four well pads developed (Fig. 1). Using the COGCC 1999–2015 spill data and values from the second scenario the likelihood of a spill is 0.44 spills/single-well pad, therefore there would likely be a spill for every three well pads developed (Fig. 1).