reducing the burden of repetitive tasks within a project ...• create python script to convert...
TRANSCRIPT
Reducing the burden of repetitive tasks within a project environment.
12/6/2019
Dr James Smith
CTO
A massive thanks to:
J U N E 2 0 1 9
Dean Murphy
Pre-sales Technical Consultant
True Enterprise RPA
Who am I?
Completed my PhD in applied Mathematics in 2015 – Painlevé equations and Orthogonal Polynomials
Lectured mathematics at the University of Kent
Joined Projecting Success in 2017
Interests:
Machine Learning Python
Mathematics
What’s in store?!
• An introduction to RPA
▪ What is RPA?
▪ What can RPA do?
▪ Why you need help?
• Some background information
• The use cases:
▪ Projecting Success – Processing employee timesheets.
▪ KIER – BIM models.
▪ Data Friends – Web scraping data.
▪ Balfour Beatty – Automating document naming conventions.
▪ Osborne – Optimising how to process delivery notes.
J U N E 2 0 1 9
Dean Murphy
Pre-sales Technical Consultant
True Enterprise RPA
Some background information
• Demonstrate the power of RPA to the community
• Test different technology stacks and IT systems to solve different business problems
• Refine our RPA roadmap
The use cases for today:
▪ Projecting Success – Processing employee timesheets.
▪ KIER – BIM models.
▪ Data Friends – Web scraping data.
▪ Balfour Beatty – Automating document naming conventions.
▪ Osborne – Optimising how to process delivery notes.
Processing employee timesheets.
Processing employee timesheets.
Processing employee timesheets.
• Approximately 40 hours a year.
• Complicated process with 3 systems working in conjunction.
• Vital that mistake are not made as people’s pay is at stake
• Easy to miss unusual activity.
The Numbers..
Processing employee timesheets.
Processing employee timesheets.
• Approximately 40 hours a year.
• Complicated process with 3 systems working in conjunction.
• Vital that mistake are not made as people’s pay is at stake
• Easy to miss unusual activity.
The Benefits…
• Reduced to 0 hours a year
• No need to login or interact with any systems
• The chance for error is reduced with the human removed.
• Python script checks to look for unusual actively. Because it’s Python, this can be as extensive as you want!
Further developments
• Adapt Python script to find more anomalies.
• Schedule the process that it can run un-assisted.
The use cases for today:
▪ Projecting Success – Processing employee timesheets.
▪ KIER – BIM models.
▪ Data Friends – Web scraping data.
▪ Balfour Beatty – Automating document naming conventions.
▪ Osborne – Optimising how to process delivery notes.
BIM Model Federation
• Log into BIMXtra
• Download updated models
• Extract these models
71 step process. Key steps are detailed below:
• Login to email to retrieve 2 factor authentication code.
• Process email to uncover 2 factor code and enter in to website.
• Extracting models to various locations for, not only "LIVE" models but also for archiving, Improving on KIER's current system.
• The model navigates through BIMXtra to the right location.
• A feature can be added to download the models that are required.
BIM Model Federation
• Open live smc file in Solibriand update models
• Save smc file to archive and "LIVE" folder
• Send email to team
• Navigate to the live SMC file• Open the file within the Solibri
software• Potentially massive loading
times have to be dealt with
• Even the automation of the emailing of the team can be completed.
• Saving two copies of the process
BIM models
• Approximately 150 hours a year (per project).
• Complicated process with 3 systems working in conjunction.
• Vital that the team is kept up to date.
The Numbers..
BIM models
BIM models
• Reduced to 0 hours a year.
• No need to login or interact with any systems.
• Process can be scheduled to run as and when you want.
The Benefits…
• Approximately 150 hours a year (per project).
• Complicated process with 3 systems working in conjunction.
• Vital that the team is kept up to date.
Further developments
• Add functionality to download more than 1 model at a time• All modified models• All models modified in the last
week/day/month
• Add functionally to choose which project to download the models for
The use cases for today:
▪ Projecting Success – Processing employee timesheets.
▪ KIER – BIM models.
▪ Data Friends – Web scraping data.
▪ Balfour Beatty – Automating document naming conventions.
▪ Osborne – Optimising how to process delivery notes.
Web scraping data
https://www.wgea.gov.au/public-reports
Web scraping data.
The Numbers..
• 5,000 records. One minute to manually download a record.
• £500 to complete this manually
• Can use Python but coding experience is needed
• Protected websites can be a problem
Web scraping data.
Web scraping data.
The Benefits…
• Now takes 1 hour (unattended)
• Now this can run in the background
• If you know what you’re doing - takes minutes to setup with UiPath studio
• Uipath can easily login and is actually a browser
• 5,000 records. One minute to manually download a record.
• £500 to complete this manually
• Can use Python but coding experience is needed
• Protected websites can be a problem
Further developments
• Schedule task to run periodically.
• Performance could be improved by removing UiPath “pauses”.
The use cases for today:
▪ Projecting Success – Processing employee timesheets.
▪ KIER – BIM models.
▪ Data Friends – Web scraping data.
▪ Balfour Beatty – Automating document naming conventions.
▪ Osborne – Optimising how to process delivery notes.
Automation ProjectWise files (models)
Automating document naming conventions
Automating document naming conventions
The Numbers..
• Time consuming naming documents manually. It takes a long time for us to find the correct Asset ID (AIMSID), FRMS and Asset Type. EA have given about 5000 asset ids and some of them share the same address and locations.
• Time wasted across all projects can quickly increases costs through lost time incurred by Balfour Beatty and Jacobs
• As document numbers are booked in bulk but are not confirmed until document is uploaded to ProjectWise there is also an issue with duplication of document numbers
Automation
Individual Project Cost Reports
Cost Reporting File – TEAM2100 – Currently reports are manually separated by cost manager and sent to QS individually for review
Automating Cost Reporting
The Numbers..
• To create each report currently takes cost manager 2hrs. QS’s then spend 2hrs reviewing.
• 4hrs = 96hrs/year/job. We have about 10-12 jobs ongoing so upwards of 960 hrs or 120 work days a year.
• The reports are cumulative so rely on the QS spotting all the differences every month and reporting on them. The reporting period is short so this is always rushed leading to misreporting costs and stress to the workforce.
Automating Cost Reporting
The use cases for today:
▪ Projecting Success – Processing employee timesheets.
▪ KIER – BIM models.
▪ Data Friends – Web scraping data.
▪ Balfour Beatty – Automating document naming conventions.
▪ Osborne – Optimising how to process delivery notes.
Some information on Osborne:
Construction company working within rail, road housing and healthcare sectors.
In 2018 Osborne had:
• 166 live projects
• Submitted 4757 orders
• Ordered 16,499 individual line items
Optimising how to process delivery notes
The Numbers..
• Approximately 2000 hours a year.
• Process costs £80,000 per year
• An on-site member has to spend time typing out the delivery note. This can be demotivating for employees.
• Errors are made when copying the delivery note. This has a HUGE knock on effect when the office try to work out what has been delivered.
•Ordered 16,499 individual line items
Optimising how to process delivery notes
VS
OCR??
On site
Off site
A web order is made
Delivery made Typed delivery note
Supplier is paid Invoice MRS MRS received
Optimising how to process delivery notes
VS
Off site
Delivery made
Supplier is paid Invoice MRS MRS received
Order confirmation
A web order is made
On site
Optimising how to process delivery notes
Web scraping data
The Benefits…
• This will now be massively reduced.
• This will now be massively reduced.
• No typing out delivery notes anymore so employees feel more valued. Improving morale.
• As order confirmations are now digital there are no errors processing this data.
• Approximately 2000 hours a year.
• Process costs £80,000 per year
• An on-site member has to spend time typing out the delivery note. This can be demotivating for employees.
• Errors are made when copying the delivery note. This has a HUGE knock on effect when the office try to work out what has been delivered.
Further developments
• Add the option to upload documents• Save time photocopying• Reduce the chance that these files are lost.
• Add functionality for parts of the delivery on backorder.
• Expand the scope of the Python script to cope with more supplier types
• Create Python script to convert invoices in to data.
• Create UiPath process that automatically compares MRS with extracted data from invoices.
Our conclusions
1. Demonstrate the power of RPA to the community
2. Test different technology stacks and IT systems to solve different business problems
3. Refine our RPA roadmap
Our conclusions
Front End Connections Backend
APIs
2. “Test different technology stacks and IT systems to solve different business problems”
Our conclusions
RPA is not just process replication, it’s more than that, we optimise your process by doing a process review so it’s more:
1. Efficient
2. Robust
3. Timely
+ Previously unforeseen benefits
(preparation for Machine Learning)
Start
Understand process
Process Review
Replicate Process?
3. “Refine our RPA roadmap”
That’s it!
Please find me on Linkedin
James Smith
Dr James Smith
CTOwww.projectingsuccess.co.uk
+44 7891805279