Once, twice, three times a maybe

Dean Palmer gets some good advice from an engineering company that has been through the pains of implementing PLM software no less than three times in the last 13 years

Alan Smith is no stranger when it comes to implementing product data management (PDM/PLM) software – he's done it three times in the last 13 years. At a PLM seminar at Warwick University last month he told the audience: "Each project has taught us volumes in terms of implementing PLM, but three things really stand out as critical to PLM's success: managing your company's expectations; making compromises in order to get the best out of the software; and you have to realise there are no shortcuts - PLM is not cheap, nor easy to implement and anything you hear to the contrary is just marketing spin." Smith is the CAD/PDM manager of Elmar, an engineering company headquartered in Aberdeen that designs and manufactures a range of wireline and completion products for the oil industry. This includes small, mass-produced, machined components through to large, high value truck-mounted masts and winch units. The company employs 300 staff across six manufacturing sites world-wide. Smith defined PLM as "a toolkit of software components from one or more vendors, configured and linked together by one or more of those vendors to increase the ability of a user to define and manage the lifecycle of his product in a (largely) digital environment." According to Smith, in 1990, when Elmar first implemented PLM (or PDM as it was called back then) the company and its design office had the following systems in place: automated workflow with electronic sign-off; vaulted storage of data; automated drawing border properties; automatic parts lists on drawings; automatic transfer of parts lists to the MRP system; and control of the lifecycle of some product data. He explained: "We use different CAD packages, different MRP systems, different hardware, different data, different properties and different QA processes. So the directors at that time wanted to invest in [CAD and business] software that would allow the business to grow whilst automating and/or eliminating as many of the repetitive processes as possible. They knew they had to throw out the drawing boards." And throw them out they did. In the Summer of 1990, Elmar evaluated and selected a 5-user CAD software license running on Sun Sparc 5's along with a business application (with unlimited licenses) that ran on an Altos 386-based server. "It took three months to evaluate the systems, three months to plan it all, three to configure it, and six months from 'go live' to make it all work," said Smith. "That meant an overall project duration of 15 months." On the plus-side, the new systems gave Elmar a high level of automation, including automatic transfer of parts and bills of material (BOMs) to the MRP system. And there was no per user license cost. And based on the number of users and number of drawings produced, Smith reckoned the productivity of his department went up by at least 156%, from 78 drawings per user per year to 200. But on the down side, there was a heavy reliance on the software supplier and the cost of growth, support and to develop extra functionality was too high. The CAD and business system was used up to 1998, by which time the software was almost obsolete. Smith explained: "There was a decreasing level of support from the vendor, the hardware was old and the software vendor had no 3D CAD software and was unable to adopt 3D CAD with its existing data management software. So we looked elsewhere." The second PLM project started in early 1996 and came about because the directors wanted to move to 3D CAD. Smith said there was no fixed budget, just a gut feeling from the directors on approximately how much they were willing to spend. "There was no budget framework to predict ROI; no criteria set to enable demonstrable success to be measured; and the team involved were all engineering personnel, there was no buy-in from other departments." Next, Smith and his team visited users, CAD software shows and spoke to a few suppliers. "We selected [Unigraphics] 3D modelling software first as the main functional criteria, but whoever we selected had to provide the PDM side too," Smith continued. "We considered only major vendors who would be unlikely to disappear and our selection criteria was based on cost, the system's ability to manage our existing 2D CAD data, and we needed a turnkey solution which included 3D CAD, PDM, hardware and integration." The system selected was for 18 2D CAD seats and four 3D CAD licenses all running on Unix, and 25 PDM users. Smith said the project this time took a total of 27 months to complete. "It took us six months to evaluate and select a vendor; three months of planning; nine months to configure the system; and nine months from 'go live' in April 1998 to make it all work," he added. As well as taking longer than expected to implement, there were other problems. Smith said the project was hindered from the start due to a poor hardware platform. "The software was so expensive that we were advised by the vendor to cut the hardware budget. In reality, this hardware was not able to deliver the performance we required. We ended up losing one man-year just waiting for design transactions to occur." Smith also referred to “constantly changing vendor account managers”, “high consultancy costs due to development and growth of the system” and “complexity of the 3D product” as further hurdles to the project. In August 2000, the company decided to launch a third PLM project. Smith said the reasons for considering upgrading or changing the PLM system were clear at the time: “Our existing software vendor’s culture and size did not match ours. The support varied depending on how tied up that supplier was with other larger customers at the time. There was also a lack of ability to see things from our perspective. That is, our expectations were not delivered.” He also cited “cost reduction” and the fact that the existing system could not model some of the company’s “fundamental” workflow processes as further proof that things had to change. Whereas PLM 1 had yielded 156% productivity gains, PLM 2 gave Elmar just a 15% gain. Smith said this was based on 41,000 drawings produced using PLM 2 by 24 users, giving a productivity of 230 drawings per year. Started initially as a feasibility study in late 2000, PLM 3 was purely an exploratory exercise within engineering. According to Smith it was a project to explore all of the options that would deliver satisfactory user performance to staff world-wide. Also, the company wanted to migrate to a common platform (NT/Windows 2000) and it needed to share product data on a global basis with other Elmar sites. The selection criteria this time around would be system performance, cost reduction and confidence in the vendor. Smith continued: “We decided to go to the market and speak to as many vendors and resellers as possible and to contacts within other engineering firms… From this, we then invited a select list of potential vendors to tender.” Elmar needed 28 3D CAD high-end PC workstations and a 50-user PLM software licence. “In the end,” explained Smith “we went to an independent PDM vendor.” The project this time took 24 months to complete, with six months for selection, three for planning, six for system configuration, and nine months for acceptance after “go live”. “This amounted to a 50% overrun on what we had planned,” said Smith. On the plus side, Smith admitted that performance in database transactions was up, a modern Windows graphical user interface went down well with users, and the flexibility and ease of internal support were better than previous systems. The only factor against the implementation (other than running over time and budget!) was the increased skill set and awareness required by Elmar staff to maintain the 3D CAD files. Smith added: “So far, PLM 3 has gone pretty well. It’s a bit early [acceptance was only in Autumn 2003] to look at ROI and productivity, but we think the signs are good.” Elmar is an example which may put off many potential PLM users, but Smith disagrees: “You have to paint the real picture of PLM to warn companies of the dangers if they don’t keep their feet on the ground. PLM is worth it in the end, but expectations and compromise are the key issues here.” Lessons learned at Elmar * ‘Out of the box’ PLM is a myth * There are no real shortcuts, just paths with more or less hassle to achieve the same goal. And configuration is always necessary * PLM isn’t cheap nor is it easy to implement Document business processes, lifecycles you want to manage, expectations (both overall and in detail) and task responsibilities * Keep staff informed along the way and anticipate the range of learning curves * Ensure you see systems doing almost exactly what you want, or be prepared for more compromises * Be prepared to move the goalposts and change your methodologies and ideals as you adapt to the new software