Determining the Magnitude of Wind Loads on Structures in Kenya according to the Eurocodes

This study provided guidance on the use of existing wind data in Kenya with the Eurocodes despite the absence of the local national annexes. The determination of wind loads in the structural design of buildings according to the Eurocode Standard KS EN 1991-1-4:2005 had several challenges. The code of practice commonly used in Kenya was CP3-Chapter V-2:1972 that used the three-second gust duration. This gust duration resulted in higher magnitudes of wind loads that ended up making the structures unnecessarily robust and uneconomical. Using the Eurocodes had the promise of achieving more economical designs because it used the 10-minute gust duration. The 10-minute gust duration resulted in typically lower magnitudes of wind loads than the three-second gust duration for the same wind speed. Kenya adopted the Eurocodes in September 2012 but had not yet developed its national annexes opting instead to use the UK National Annexes. The UK National Annexes were applicable to Kenya in some scenarios but not in others such as wind loading. The lack of the Kenya National Annexes led to difficulties in the adoption of the Eurocodes. This paper outlined a procedure in which the existing wind data given in three-second gusts could be converted to 10-minute wind speeds. Once converted, the method described in the UK National Annex could then be followed selectively to determine the wind load on a structure. Lastly, the paper recommended that wind data collected from 1977 to 2021 by the Kenya Meteorological Department be incorporated to the development of the wind map for the Kenya National Annex to KS EN 1991-1-4:2005.


Background
Wind actions are one of the loads that are of great significance in determining the stability of structures. There are two main methods of quantifying wind speeds: sustained wind speeds and wind gusts. Sustained winds speeds are the steady or average wind speeds over a duration of time usually over a 10-minute duration. The United States of America (USA) preferred a one-minute duration for sustained wind speeds. Wind gusts are sudden increases in wind speeds that are transient and last for less than 20 seconds [1]. Wind gusts are usually about 30% stronger than sustained wind speeds [2]. Wind gusts are quite variable and depend on different dynamics such as location and terrain. Sustained wind speeds give a more accurate idea of how steady the wind speed is as opposed to wind gusts.

Problem Statement
Developers in Kenya avoided the services of professional engineers mainly because of the following reasons related to cost: engineering services were perceived to be expensive in comparison to services offered by experienced technicians; and engineering specifications were deemed to be unnecessarily conservative, resulting in additional expenses. This paper focused on the magnitude of wind actions/loads applied to structures in Kenya in line with the local regulations. The magnitude of the loads at the design stage contribute significantly to the cost of the construction project. The practice in Kenya was to use the three-second wind gusts which are higher in magnitude than the sustained wind speeds. Kenya adopted the Eurocodes on 14 th September 2012 via Gazette Notice No. 13048 [3]. The Eurocodes were a set of harmonised European standards for the design of buildings and civil engineering structures. The specific Eurocode standard for wind loading was KS EN 1991-1-4:2005 -Eurocode 1: Actions of Structures -Part 1-4: General actions -Wind actions. KS EN 1991-1-4:2005 specified that the10-minute wind speed was to be used [4]. Wind data available locally was provided by the standardisation body, Kenya Bureau of Standards (KEBS), and the weather monitoring service, Kenya Meteorological Department (KMD) and were in threesecond gusts. To a structural engineer, the main problem was inadequate guidance on how to convert threesecond gusts to 10-minute wind speeds. The objective of this paper was to provide a procedure on how to determine the magnitude of wind loads on structures in Kenya when designing to the Eurocodes.

Justification of the Study
The main justification of this study was that the Eurocodes were adopted in Kenya without the relevant corresponding National Annexes. The study demonstrated how to modify existing wind speeds provided in three-second gusts by both KEBS and KMD, to 10-minute wind speeds.

Scope
The study was limited to using existing data published by KEBS in 1977 [5]. However, due to the effects of climate change and other human activities such as deforestation, there was a high likelihood that this wind data was outdated by 2021. Unfortunately, the wind data published in 1977 was still in use in Kenya by January 2021.

Methodology
This paper was formulated through a desk study that looked into the process of determining the magnitude of wind loads on structures in Kenya and compared it with the process used in the United Kingdom of Great Britain and Northern Ireland (UK). The comparison with the UK was made because Kenya's engineering practice was heavily influenced by the UK's practice. The study relied on data from KEBS and the author's experience as a practising structural engineer in Kenya. Emphasis was put on areas where the UK practice did not apply to Kenya.

The Practice in Kenya
The practice in Kenya was based on a previous design code of the UK. That design code was CP3-Chapter V-2:1972 -Code of Basic Data for the Design of Buildings [6]. Ideally, CP3-Chapter V-2:1972 should only have been used in the UK but its application found its way to Kenya. Kenya's engineers ignored the UK wind map in CP3-Chapter V-2:1972 and instead used Table 1 Table 2 had not been published officially by any authority [7], nonetheless, the table found widespread use among engineers in Kenya. The wind speeds in Table 2 were also based on a three-second gust duration.
Once the designer made the decision on the wind speed to use at a site using either Table 1 or Table 2, the procedure in the CP3-Chapter V-2:1972 was then followed to determine the magnitude of the wind load on the structure.
Information also worth noting was that KMD released wind speeds that slightly differed from the data published by KEBS [5]. These wind speeds were also given in three-second gusts and were presented in Table  3. Table 3 shows six towns in Kenya with their corresponding wind speeds over 25-year, 50-year and 100-year return periods.  Table 3: Wind speeds for some towns in Kenya [8]. Table 1 and Table 2 did not indicate the probability of exceedance even though the 50-year return period was assumed. The 50-year return period was typically used for designing buildings [7]. A 50-year return period means that there was a 0.02 probability of exceedance, equivalent to a 50 year mean recurrence interval [9]. The wind speed in Table 1 was comparable to the wind speed on the highlighted column of Table 3 showing wind speeds for the 50-year return period.

Eurocode Adoption in Kenya
The Eurocodes had Nationally Determined Parameters (NDPs) that every country had to provide for the effective use of the codes. These NDPs were typically published in separate documents from the actual Eurocode standards. Those separate documents were referred to as National Annexes. Unfortunately, Kenya had not yet developed such annexes. Engineers continued to use Table 1 when determining the wind speeds at a site. Table 1 had wind speeds in three-second wind gusts. This gust period was used because of the response time of mechanical anemometers. The response time for anemometers was usually one to three seconds and was a function of the inertia of the system [12]. This response time was the main reason that several other countries, such as the UK, US and Australia, previously used the three-second gust duration when recording wind speeds [12]. The Eurocodes utilised 10-minute wind speeds [4]. A 10-minute gust duration is likely to be of lower speed than a three-second gust duration. This lower speed is due to the fact that over a longer period, mean velocities are lower since there is a high chance that a peak is cancelled out in the duration of the recording. Since the 10-minute gust is lower than the three-second gust, using the Eurocodes rather than using CP3-Chapter V-2:1972, could lead to the justification of more economical designs. The issue of conservativeness versus design efficiency and rationalising the use of resources was still a core question in the construction industry. Most building codes and standards, including the Eurocodes, have sided with safety and opted for more conservative designs. The gain in economy achieved by KS EN 1991-1-4:2005 did not compromise the safety of buildings. The decision that guided the reduction in the magnitude of the wind loads was based on science and research [13].

UK National Annexes
The Government of Kenya declared the Eurocodes as its national standards although Kenya's standardisation body, KEBS, had not yet developed the relevant annexes [3]. An article in The Kenya Engineer, the journal of the Institution of Engineers of Kenya (IEK), stated that the UK National Annexes (UK NA) were to be used with the adopted Eurocodes [14]. That article implicitly described the UK National Annexes as one single document. That implicit description was not accurate. There were several annexes to be used with the Eurocodes. Each specific part of the Eurocode was accompanied by its corresponding relevant annex. Statements made by Jane Maina, the manager of the Engineering Division (Civil and Mechanical engineering standards) of KEBS and Professor Stanley Shitote, the Chairman of the National Implementation Committee on the Eurocodes (NICE) in Kenya, confirmed that the Eurocodes were to be used in conjunction with the UK NA for the structural design of buildings civil engineering structures in Kenya [15]. The use of the UK NA, rather than the development of the Kenya NA, might have been advantageous to Kenya as it avoided duplication of efforts. Also, the resources that would have been used in repeat studies were diverted to the training of Kenya's engineers on the adoption and use of the Eurocodes.
The UK NA were applicable in Kenya in some situations but were not applicable in other scenarios such as that of temperature, wind or snow. Kenya and the UK had different climates. The local conditions of the two countries were quite different. The wind speed in the UK was usually read directly from a wind map in the UK NA [16]. The value of the wind speed read off from the wind map was then modified based on the season, wind direction and site altitude. The basic velocity pressure b , was then calculated using Eqn (4.10) in KS EN 1991-1-4:2005. Eqn (4.10) was reproduced below as Expression (1): where: -Density of air. b -Modified wind speed based on the season, wind direction and site altitude.
After that, the basic velocity pressure b , was multiplied by the exposure factor e ( ), as prescribed by Eqn (NA.3a) of the UK NA. Eqn (NA.3a), reproduced here as Expression (2), overrode Eqn (4.8) of KS EN 1991-1-4:2005. Eqn (4.8) was reproduced later in this paper as Expression (4). The product of the exposure factor ( ) and the basic velocity pressure b resulted in the peak velocity pressure p . This peak velocity pressure was the magnitude of the wind load that engineers used to design the structure. , p = e ( ) × b (2) where: e ( ) -Exposure factor.
b -Basic velocity pressure The exposure factor e ( ) was specific to the UK and was a function of the distance to sea. The use of the exposure factor e ( ) in Kenya would not be correct.
Arguments against the use of the exposure factor ( ) of the UK NA in Kenya 1. The UK NA was based on the fact that the UK was generally at a low-lying level with an average altitude of approximately 162m above sea level. On the other hand, Kenya had an average altitude of 762m above sea level [17]; significantly higher than the average altitude of the UK. In addition to that, several areas in Kenya were highlands with a mean sea level of 1500m above sea level. 2. The UK wind map was based on a reference datum, the sea, hence the need for an altitude correction factor alt proposed in the UK NA whereas the KS 02-26:1977 Table 1 gave speed at 10m above the ground at a particular location [5]. The data in Table  1 was not based on a reference datum and there was therefore no need to apply the altitude correction factor alt in Kenya. 3. The UK was an island and experienced marine climate throughout the entire country. The distance to the sea was therefore essential when determining the wind speed at a site. In contrast, Kenya was continental with significant geographical formations that influenced the wind pattern of that locality; consequently, the distance to sea was not the main factor affecting the wind speed. 4. The UK NA, as stated in Clause NA 2.18, used an air density value of 1.226kg/m 3 [16] whereas this study found no predetermined value for the density of air in Kenya. KS EN 1991-1-4:2005 recommended a value of 1.25kg/m 3 [4].

Other Eurocode Challenges
Kenya adopted all the Eurocodes, including the Snow Eurocode [3]. Kenya was at the equator and experienced tropical climate. The country hardly experienced any snowfall except around the peaks of Mt. Kenya that had an altitude of approximately 5,200 metres above sea level. These facts did not mean that KEBS erred in its adoption of the Snow Eurocode KS EN 1991-1-3:2004 [18]. Some parts of Kenya received hail which took several days to thaw [19]. Kericho, Laikipia, Nakuru, and Nyandarua counties had records of experiencing significant hail. Hail was covered in the Snow Eurocode KS EN 1991-1-3:2004.
A different Eurocode, KS EN 1991-1-5:2010, was the standard used for the determination of thermal actions on structures [20]. The temperature map for Kenya was also not yet been developed. The absence of this temperature map hampered Eurocode adoption in Kenya and made the determination of thermal loads, which was essential in the design of bridges, a challenging task.

Kenya Wind Map
There was no wind map used in the design of structures in Kenya, instead, the values from Table 1 were used to come with a wind map. The data given in KS 02-26:1977 showed points and areas with the same wind speed. A map of these wind speeds was modelled and isotachs were drawn as shown in Figure 1. Isotachs are lines on a map with equal wind speeds [21]. The jurisdiction of the map was as per the Constitution of Kenya, 2010 [22]. Figure 1 showed a proposed wind map for Kenya. This map used the data published in 1977. It was essential to review the wind speeds in KS 02-26:1977. KMD had been collecting data on wind speed from 1977 to 2021; it was vital that this collected wind data be incorporated to the development of Kenya's wind map.

Gust Duration Conversion
Based on statistics, the shorter the wind gust period, the higher the velocity. This fact did not imply less safety since a collapse due to wind would most likely occur because of a sustained wind speed of over 10 minutes and not a wind gust of three seconds. The wind gust duration was the main difference between CP3-Chapter V-2:1972 and KS EN 1991-1-4:2005. A Durst curve, shown in Figure 2, was employed to convert the threesecond gust to the 10-minute wind speed. A Durst curve is a statistical correlation between wind speeds over different wind speed durations [23].

Interpretation
Kenya's method of determination of wind loads on structures deviated from the method presented in the UK NA. The wind duration in Kenya should be converted from 3 to 600 . The peak velocity pressure should be calculated using Expression (4), Eqn (4.8) of KS EN 1991-1-4:2005 [4]. The exposure factor e ( ) of Eqn (NA.3a) as proposed in the UK NA cannot apply in Kenya as argued previously in this paper.

Integration Conversion Procedure
The Durst curve was used to convert the wind speeds given in KS 02-26:1977 The values 1.52 and 1.06 are read off from the Durst Curve in Figure 2. This conversion was replicated in experimental studies; for example, when the UK updated their wind map from BS6399-2:1997 ( 3600 ) to the Eurocode BS EN 1991-1-4:2005 ( 600 ) it was found that the 10-minute wind speeds were 6% higher than mean hourly speeds [9]. That finding reinforced the accuracy and reliability of using the Durst Curve. Therefore, to convert the values of the wind speed from However, because of the three errors in Table 1 highlighted previously, the values in Table 2 are preferred in the determination of wind loads on structures. The wind speeds from Table 2 were converted from three-second gusts to 10-minute wind speeds using Expression (3), and the results were presented in Table 5.   Table 5 showed five geographical regions in Kenya and their corresponding 10-minute wind speeds. It was difficult to generate isotachs with the data from Table 5 because it specified wind speed for large geographical areas and not single point locations like towns, as had been specified in Table 1. The lack of data at specific points made the generation of isotachs difficult. A wind map was therefore not developed. The data in Table 5 could suffice for the design of buildings in Kenya.
Once the wind speed was determined from  (4), the procedure in the UK NA could then be followed to determine the wind loads on the structure or parts of the structure.

Implication
The use of Expression (4) required the determination of the mean wind velocity m ( ) and turbulence intensity v ( ). The mean wind velocity at a height z m ( ) depended on the terrain roughness and orography. Turbulence intensity v ( ) depended on the height above the ground level and ground roughness of the terrain upwind of the structure in the wind direction considered. The values of m ( ) and v ( ) are calculated from values read off from  Table 4.1 and was therefore insufficient in the determination of the mean wind velocity m ( ) and turbulence intensity v ( ). Engineers in Kenya could therefore ignore the UK NA in the determination of the peak velocity pressure and adopt the recommendations of KS EN 1991-1-4:2005.
The value for Nairobi City's wind speed of 19.5m/s in Table 5 was similar to the value found in a separate study where wind data from eight weather stations around Nairobi City County in Kenya, was collected and analysed [7]. That study found that the wind speed 600 , in Nairobi City County varied from 18m/s to 22m/s. That finding verified the conversion procedure of Expression (3) proposed in this paper. Since most structural engineers in Nairobi used a wind speed of 28m/s when determining wind loads on structures [24], adopting a wind speed of 19.5m/s instead of 28m/s resulted in a reduction of the magnitude of wind loads. This reduction may result in cost savings for building projects.
In summary, even though the UK NA could override KS EN 1991-1-4:2005, this study recommended that: 1. The UK wind map was be ignored in Kenya. 2. The altitude correction factor alt in the UK NA was not to be used. 3. Eqn (NA.3a) should be ignored and Eqn Except for these five issues, the UK NA could be used in Kenya effectively for the determination of wind loads in the structural design of buildings. Others features that affect wind loading stated in the UK NA such as building shape, roof pitch, terrain shielding and building dimensions are appropriate for use in Kenya.

Conclusion
It was possible to use the Eurocodes for the structural design of buildings in Kenya before KEBS finalised on the development of the Kenya National Annexes. A combination of the local table values from Table 1 of KS 02-26:1977, the conversion procedure shown in this paper, and the UK NA could be used effectively to determine the wind loads on structures when designing to the Eurocodes in Kenya.