## You are currently viewing: Articles

### Logistics of personnel movement by elevator

During a construction project time is money, and that includes time waiting for elevators to the workplace.

**JAVIER VAZQUEZ-ESPARRAGOZA and GIOVANNI PUCCINI**

KBRKBR

**Viewed :**184

**Article Summary**

This article describes a solution using discrete simulation modelling of construction personnel arriving to work and going to different floors of a building facility within a production plant during a peak construction period.

Discrete simulation modelling has been an integral tool for KBR, not only for the design and operational analysis of production facilities but also for traffic and logistics planning during construction phases. This type of modelling represents random events over time to capture detailed variations of vehicles such as cars/buses/trucks/railcars or individual people movements. These simulation studies investigate alternatives to find cost-effective improvement measures that move people and freight efficiently throughout the network without doing costly experiments in the field.

The following sections describe a specific construction personnel logistics study. In this case, it is a project consisting of a new facility designed to produce 500000 t/y of bimodal polyethylene. However, the principles apply to equivalent refinery construction and revamp projects. At the construction site, one of the units, the polyethylene production plant, faced the need to move up to 332 construction people with just two temporary elevators for the 20-floor building under construction. It is important to clarify that this logistics challenge occurred during the construction phase only because the two temporary elevators did not have a computer algorithm or controller to manage the movement and load of the elevators as in the case of commercial buildings.

The main objective of the study was to find out if the two available temporary elevators were enough; and if so, what would be their optimum arrangement to distribute people once they arrive together in the mornings and go to different floors throughout the 20-storey structure. The general solution to this problem is to calculate accurately the true capacity of the elevators by simulating the expected operations using actual equipment specifications and project conditions. From there, several scenarios are simulated in order to optimise operations among often conflicting variables. For example, an underestimated elevator capacity causes delays in daily construction and thus impacts productivity during construction. On the contrary, an overestimated elevator capacity or inefficient people operations leads to unnecessary investment in elevators. Both conditions represent unnecessary costs and thus the final solution looks for an optimum balance between capacity and delays. The proposed solution also takes into account likely uncertain events and possible variations in schedule specific to this construction project in order to form a resilient solution.

Consequently, questions needed to be addressed in this study included the following:

• What is the minimum number and size of elevators needed to support the anticipated construction operations?

• What is the optimum elevator arrangement to move construction people efficiently?

• Will there be any bottleneck in the logistics operations in terms of unacceptable queuing or delays?

• How to mitigate the potential risks subject to the construction phase?

The following procedure is used to determine the statistical distribution values:

1. Run a simulation model for the given arriving and processing time period, for instance one hour at the beginning of the construction shift

2. Repeat the simulation for a given number of days (or replications)

3. Obtain average values from the overall simulation runs

4. Obtain averages from individual replications to derive 95% confidence levels.

The model procedure uses discrete distributions to model time events, for example, if the average entrance time of one person to the elevator is one minute, the model would use a triangular distribution (for instance, TRIAG [0.5, 1.0, 1.5] minutes) to model the random times of the event. With this technique, the model takes into account the time variability of the event, from the time slower than the average (0.5 min) to the time faster than the average (1.5 min), and runs the simulation 20 times before collecting statistics. The simulation model is constructed such that it can be easily reused for ‘fine-tuning’ to accommodate future development changes such as additional elevators or arriving personnel. Also one can add a random event to cover the ‘if’ situations; for example, what would happen if one elevator needs repair time or is stopped for some period of time during the day.

In order to find a solution, a simulation model was built by using a commercial discrete simulation program and running several scenarios with some options for the movement of the personnel as defined by the need of the employees to reach certain floors and to reduce the waiting time to get into the elevators.

Table 1 shows the number of people expected per day during seven months of construction for the 20-storey facility.

The base case scenario of the simulation is for the month of April when the number of workers arriving at the elevator is expected to peak (332 according to Table 1). These employees arrive within a one-hour span (6:00am to 7:00am) and from there go to different stories of the structure to work for a complete shift (8-10 hours).

The first question is to find out if the two temporary elevators available would satisfy the highest elevator demand imposed by the base conditions above.

The following are the specific elevator data parameters collected for this study:

a. The elevators were to be placed outside the steel structure of the building

b. Building and surroundings would have enough room for a general standard length elevator. A full length car would be rated for a maximum of 30 people, depending on the size/weight of the passengers and any tools or material they may be carrying

c. The elevators would be rated for maximum 6000 lb capacity

d. The doors operate manually so open/close times can vary depending on the operator. It should take no longer than 3-4 seconds to open or close the elevator doors

e. Building has a table top second level where some of the workers access through the stairs from ground level

f. Not including start-up travel, ground level to the table top would take approximately seven seconds

g. Non-stop elevator travel times from ground level to level 20 is about 63 seconds

h. For up or down travel distances of 100 ft or less, the average elevator speed is about 150 ft/min

i. For distances above 100ft, the average speed doubles to about 300 ft/min

j. Notice that if an elevator stops every one or two floors, it will not likely reach the maximum travel speed of 300 ft/min

k. Elevator relative elevation: ground (Level 1) = 100ft; table top (Level 2) = 133ft; last floor served (Level 20) = 330ft

l. People arrive between 6:am and 7:am. For example, during the month of April, a person will arrive on average every 60/332 = 0.1807 minutes

m. Most of the event times could be simulated with a discrete distribution function; for example, the arrival of workers to the building was modelled as EXPO (0.1807) minutes, and the time to walk to the elevators area as TRIA (15,45,90) seconds.

Discrete simulation modelling has been an integral tool for KBR, not only for the design and operational analysis of production facilities but also for traffic and logistics planning during construction phases. This type of modelling represents random events over time to capture detailed variations of vehicles such as cars/buses/trucks/railcars or individual people movements. These simulation studies investigate alternatives to find cost-effective improvement measures that move people and freight efficiently throughout the network without doing costly experiments in the field.

The following sections describe a specific construction personnel logistics study. In this case, it is a project consisting of a new facility designed to produce 500000 t/y of bimodal polyethylene. However, the principles apply to equivalent refinery construction and revamp projects. At the construction site, one of the units, the polyethylene production plant, faced the need to move up to 332 construction people with just two temporary elevators for the 20-floor building under construction. It is important to clarify that this logistics challenge occurred during the construction phase only because the two temporary elevators did not have a computer algorithm or controller to manage the movement and load of the elevators as in the case of commercial buildings.

The main objective of the study was to find out if the two available temporary elevators were enough; and if so, what would be their optimum arrangement to distribute people once they arrive together in the mornings and go to different floors throughout the 20-storey structure. The general solution to this problem is to calculate accurately the true capacity of the elevators by simulating the expected operations using actual equipment specifications and project conditions. From there, several scenarios are simulated in order to optimise operations among often conflicting variables. For example, an underestimated elevator capacity causes delays in daily construction and thus impacts productivity during construction. On the contrary, an overestimated elevator capacity or inefficient people operations leads to unnecessary investment in elevators. Both conditions represent unnecessary costs and thus the final solution looks for an optimum balance between capacity and delays. The proposed solution also takes into account likely uncertain events and possible variations in schedule specific to this construction project in order to form a resilient solution.

Consequently, questions needed to be addressed in this study included the following:

• What is the minimum number and size of elevators needed to support the anticipated construction operations?

• What is the optimum elevator arrangement to move construction people efficiently?

• Will there be any bottleneck in the logistics operations in terms of unacceptable queuing or delays?

• How to mitigate the potential risks subject to the construction phase?

The following procedure is used to determine the statistical distribution values:

1. Run a simulation model for the given arriving and processing time period, for instance one hour at the beginning of the construction shift

2. Repeat the simulation for a given number of days (or replications)

3. Obtain average values from the overall simulation runs

4. Obtain averages from individual replications to derive 95% confidence levels.

The model procedure uses discrete distributions to model time events, for example, if the average entrance time of one person to the elevator is one minute, the model would use a triangular distribution (for instance, TRIAG [0.5, 1.0, 1.5] minutes) to model the random times of the event. With this technique, the model takes into account the time variability of the event, from the time slower than the average (0.5 min) to the time faster than the average (1.5 min), and runs the simulation 20 times before collecting statistics. The simulation model is constructed such that it can be easily reused for ‘fine-tuning’ to accommodate future development changes such as additional elevators or arriving personnel. Also one can add a random event to cover the ‘if’ situations; for example, what would happen if one elevator needs repair time or is stopped for some period of time during the day.

In order to find a solution, a simulation model was built by using a commercial discrete simulation program and running several scenarios with some options for the movement of the personnel as defined by the need of the employees to reach certain floors and to reduce the waiting time to get into the elevators.

**Logistics situation of elevators**Table 1 shows the number of people expected per day during seven months of construction for the 20-storey facility.

The base case scenario of the simulation is for the month of April when the number of workers arriving at the elevator is expected to peak (332 according to Table 1). These employees arrive within a one-hour span (6:00am to 7:00am) and from there go to different stories of the structure to work for a complete shift (8-10 hours).

The first question is to find out if the two temporary elevators available would satisfy the highest elevator demand imposed by the base conditions above.

The following are the specific elevator data parameters collected for this study:

a. The elevators were to be placed outside the steel structure of the building

b. Building and surroundings would have enough room for a general standard length elevator. A full length car would be rated for a maximum of 30 people, depending on the size/weight of the passengers and any tools or material they may be carrying

c. The elevators would be rated for maximum 6000 lb capacity

d. The doors operate manually so open/close times can vary depending on the operator. It should take no longer than 3-4 seconds to open or close the elevator doors

e. Building has a table top second level where some of the workers access through the stairs from ground level

f. Not including start-up travel, ground level to the table top would take approximately seven seconds

g. Non-stop elevator travel times from ground level to level 20 is about 63 seconds

h. For up or down travel distances of 100 ft or less, the average elevator speed is about 150 ft/min

i. For distances above 100ft, the average speed doubles to about 300 ft/min

j. Notice that if an elevator stops every one or two floors, it will not likely reach the maximum travel speed of 300 ft/min

k. Elevator relative elevation: ground (Level 1) = 100ft; table top (Level 2) = 133ft; last floor served (Level 20) = 330ft

l. People arrive between 6:am and 7:am. For example, during the month of April, a person will arrive on average every 60/332 = 0.1807 minutes

m. Most of the event times could be simulated with a discrete distribution function; for example, the arrival of workers to the building was modelled as EXPO (0.1807) minutes, and the time to walk to the elevators area as TRIA (15,45,90) seconds.

- Categories :
- Process Modelling and Simulation

**Current Rating :**1